# total-recall - Doramagic AI Context Pack > Positioning: a pre-install experience and judgment asset. It helps the host AI get off to a good start, but it does not mean the project has already been installed, run, or validated. ## Sufficiency Principle - **Sufficiency over compression**: The AI Context Pack should be sufficient for the host AI to understand the project's value, capability boundaries, entrypoints, risks, and evidence sources before starting work; it may be layered, but it does not aim for the shortest possible summary. - **Compression policy**: Compress only noise and duplication, never context that affects judgment or the quality of the work. ## How the Host AI Should Use This You are reading the AI Context Pack that Doramagic compiled for total-recall. Treat it as pre-work context: help the user understand who it fits, what it can do, how to start, what must be verified after install, and where the risks are. Do not claim that you have already installed, run, or executed the target project. ## Claim Consumption Rules - **Fact source**: Repo Evidence + Claim/Evidence Graph; the Human Wiki only supplies salience, terminology, and narrative structure. - **Minimum status for a fact**: `supported` - `supported`: May be used as a project fact, but the answer must cite the claim_id and evidence path. - `weak`: Usable only as a low-confidence lead; the user must be asked to keep verifying. - `inferred`: Usable only for risk notes or open questions; must not be packaged as a project fact. - `unverified`: Must not be used as fact; state clearly that evidence is insufficient. - `contradicted`: Must show the conflicting sources and must not force a single version on the user's behalf. ## Who It Fits Best - **Developers already using host AIs such as Claude/Codex/Cursor/Gemini**: The README or plugin config mentions multiple host AIs. Evidence: `README.md` Claim: `clm_0004` supported 0.86 - **Users who want to bring professional workflows into a host AI**: The repo contains Skill documents. Evidence: `skills/llm-setup/SKILL.md`, `skills/recall/SKILL.md`, `skills/speak-like-operator/SKILL.md` Claim: `clm_0005` supported 0.86 ## What It Can Do - **AI Skill / Agent Instruction Asset Library** (Previewable before install): The project contains Skill or Agent instruction files that a host AI can read, useful for bringing professional workflows into hosts like Claude, Codex, or Cursor. Evidence: `skills/llm-setup/SKILL.md`, `skills/recall/SKILL.md`, `skills/speak-like-operator/SKILL.md` Claim: `clm_0001` supported 0.86 - **Multi-Host Install and Distribution** (Verify after install): The project contains plugin or marketplace configuration, indicating it targets install and distribution across one or more AI hosts. Evidence: `.claude-plugin/marketplace.json`, `.claude-plugin/plugin.json` Claim: `clm_0002` supported 0.86 - **Command-Line Startup or Install Flow** (Verify after install): The project documentation contains runnable commands; real use requires running them in a local or host environment. Evidence: `README.md` Claim: `clm_0003` supported 0.86 ## How to Start - `/plugin marketplace add 88plug/claude-code-plugins` Evidence: `README.md` Claim: `clm_0006` supported 0.86 - `/plugin install total-recall@88plug` Evidence: `README.md` Claim: `clm_0007` supported 0.86 - `git clone https://github.com/88plug/total-recall.git` Evidence: `README.md` Claim: `clm_0008` supported 0.86 ## Continue-or-Stop Decision Card - **Current recommendation**: Sandbox trial only - **Why**: The project has signals of install commands, host configuration, or local writes; do not go straight into your primary environment—trial it in isolation first. ### 30-Second Read - **What to do now**: Sandbox trial only - **Minimum safe next step**: Run Prompt Preview first; if you still want to install, trial only in an isolated environment - **Do not trust yet**: Real output quality cannot be trusted before install. - **Continuing will touch**: Command execution, Host AI configuration, Local environment or project files ### What You Can Trust Now - **Target-audience signal: Developers already using host AIs such as Claude/Codex/Cursor/Gemini** (supported): Backed by a supported claim or project evidence, but that still is not the same as real install results. Evidence: `README.md` Claim: `clm_0004` supported 0.86 - **Target-audience signal: Users who want to bring professional workflows into a host AI** (supported): Backed by a supported claim or project evidence, but that still is not the same as real install results. Evidence: `skills/llm-setup/SKILL.md`, `skills/recall/SKILL.md`, `skills/speak-like-operator/SKILL.md` Claim: `clm_0005` supported 0.86 - **Capability exists: AI Skill / Agent Instruction Asset Library** (supported): You can trust that the project contains signals of this capability; whether it fits your specific task still needs trial or after-install verification. Evidence: `skills/llm-setup/SKILL.md`, `skills/recall/SKILL.md`, `skills/speak-like-operator/SKILL.md` Claim: `clm_0001` supported 0.86 - **Capability exists: Multi-Host Install and Distribution** (supported): You can trust that the project contains signals of this capability; whether it fits your specific task still needs trial or after-install verification. Evidence: `.claude-plugin/marketplace.json`, `.claude-plugin/plugin.json` Claim: `clm_0002` supported 0.86 - **Capability exists: Command-Line Startup or Install Flow** (supported): You can trust that the project contains signals of this capability; whether it fits your specific task still needs trial or after-install verification. Evidence: `README.md` Claim: `clm_0003` supported 0.86 - **There are Quick Start / install-command signals** (supported): You can trust that the docs mention a startup or install entrypoint; do not run it directly in your primary environment because of that. Evidence: `README.md` Claim: `clm_0006` supported 0.86 ### What You Cannot Trust Yet - **Real output quality cannot be trusted before install.** (unverified): Prompt Preview can only show how it guides you; it cannot prove result quality in the real project. - **Host AI version compatibility cannot be trusted before install.** (unverified): Host loading rules and version differences across Claude, Cursor, Codex, Gemini, and others must be verified in a real environment. - **That it will not pollute your existing host AI's behavior cannot be trusted directly.** (inferred): Skill, plugin, and AGENTS/CLAUDE/GEMINI instructions may change the host AI's default behavior. Evidence: `.claude-plugin/marketplace.json`, `.claude-plugin/plugin.json`, `CLAUDE.md`, `skills/llm-setup/SKILL.md` et al. - **Safe rollback cannot be assumed by default.** (unverified): Unless the project clearly provides uninstall and recovery instructions, verify in an isolated environment first. - **After a real install, is it compatible with the user's current host AI version?** (unverified): Compatibility can only be verified in the actual host environment. Evidence: `.claude-plugin/marketplace.json`, `.claude-plugin/plugin.json` - **Does the project's output quality meet the user's specific task?** (unverified): The pre-install preview can only show flow and boundaries; it cannot replace real evaluation. - **Do the install commands require network access, permissions, or global writes?** (unverified): This affects install risk in both enterprise and personal environments. Evidence: `README.md` ### What Continuing Will Touch - **Command execution**: Package managers, network downloads, the local plugin directory, project config, or the user's home directory. Why: Running the very first command can already change your environment; decide whether it is worth running first. Evidence: `README.md` - **Host AI configuration**: The plugin, Skill, or rule-loading config of hosts like Claude/Codex/Cursor/Gemini/OpenCode. Why: Host configuration changes how the AI works afterward and may conflict with the user's existing rules. Evidence: `.claude-plugin/marketplace.json`, `.claude-plugin/plugin.json`, `CLAUDE.md`, `skills/llm-setup/SKILL.md` et al. - **Local environment or project files**: Install results, plugin caches, project config, or local dependency directories. Why: The write scope and rollback path cannot be proven before install and need isolated verification. Evidence: `.claude-plugin/marketplace.json`, `.claude-plugin/plugin.json`, `README.md` - **Host AI context**: The AI Context Pack, Prompt Preview, Skill routing, risk rules, and project facts. Why: Importing context affects the host AI's later judgment, so avoid packaging unverified items as facts. ### Minimum Safe Next Steps - **Run Prompt Preview first**: Use a pre-install interactive trial to judge whether the way of working fits; it needs no authorization or environment change. (applies when: Applies to any project, especially when output quality is unknown.) - **Trial-install only in an isolated directory or a test account**: Avoid letting install commands pollute your primary host AI, real projects, or home directory. (applies when: When there are signals of command execution, plugin config, or local writes.) - **Back up your host AI configuration first**: Skill, plugin, and rule files may change the default behavior of Claude/Cursor/Codex. (applies when: When there is a plugin manifest, a Skill, or a host rule entrypoint.) - **After install, verify just one minimal task**: Verify loading, compatibility, output quality, and rollback first, then decide whether to use it deeply. (applies when: When moving from a trial into a real workflow.) ### Exit Plan - **Preserve the pre-install state**: Record the original host config and project state so you can later judge whether it is recoverable. - **Be ready to remove the host plugin / Skill / rule entrypoint**: If behavior is off after the trial install, you can restore the host AI to its pre-trial state. - **Record the install commands and written paths**: Without clear uninstall instructions, you at least need to know which directories or configs to clean up manually. - **If there is no rollback path, do not enter your primary environment**: No rollback is a blocker before continuing; do not proceed on trust or luck. ## What Can Only Be Previewed - Explain who the project fits and what it can do - Demonstrate a typical conversation flow based on project docs - Help the user decide whether it is worth installing or researching further ## What Must Be Verified After Install - Actually installing the Skill, plugin, or CLI - Running scripts, modifying local files, or accessing external services - Verifying real output quality, performance, and compatibility ## Boundary & Risk Decision Card - **Mistaking the pre-install preview for a real run**: The user may overestimate how much configuration, permission, and compatibility verification the project has already done. Mitigation: Clearly separate prompt_preview_can_do from runtime_required. Claim: `clm_0009` inferred 0.45 - **Host AI plugin or Skill rule conflicts**: New rules may change how the user's existing host AI behaves. Mitigation: Inspect the plugin manifest and Skill files before installing, and test in isolation if needed. Evidence: `.claude-plugin/marketplace.json`, `.claude-plugin/plugin.json` Claim: `clm_0010` supported 0.86 - **Command execution will modify the local environment**: Install commands may write to the user's home directory, the host plugin directory, or project configuration. Mitigation: Run in an isolated environment or a test account first. Evidence: `README.md` Claim: `clm_0011` supported 0.86 - **To confirm**: After a real install, is it compatible with the user's current host AI version?. Why: Compatibility can only be verified in the actual host environment. - **To confirm**: Does the project's output quality meet the user's specific task?. Why: The pre-install preview can only show flow and boundaries; it cannot replace real evaluation. - **To confirm**: Do the install commands require network access, permissions, or global writes?. Why: This affects install risk in both enterprise and personal environments. ## Pre-Work Working Context ### Loading Order - First read how_to_use.host_ai_instruction to establish the boundaries of this pre-install judgment asset. - Read claim_graph_summary to confirm facts come from the Claim/Evidence Graph, not the Human Wiki narrative. - Then read intended_users, capabilities, and quick_start_candidates to judge whether the user is a match. - When you need to carry out a concrete task, check role_skill_index first, then evidence_index. - For real install, file modification, network access, performance, or compatibility questions, turn to risk_card and boundaries.runtime_required. ### Task Routes - **AI Skill / Agent Instruction Asset Library**: Use role_skill_index / evidence_index to help the user pick a usable role, Skill, or workflow first. Boundary: Can be experienced via a pre-install Prompt. Evidence: `skills/llm-setup/SKILL.md`, `skills/recall/SKILL.md`, `skills/speak-like-operator/SKILL.md` Claim: `clm_0001` supported 0.86 - **Multi-Host Install and Distribution**: State that this is an after-install capability first, then give a pre-install checklist. Boundary: Must be verified after a real install or run. Evidence: `.claude-plugin/marketplace.json`, `.claude-plugin/plugin.json` Claim: `clm_0002` supported 0.86 - **Command-Line Startup or Install Flow**: State that this is an after-install capability first, then give a pre-install checklist. Boundary: Must be verified after a real install or run. Evidence: `README.md` Claim: `clm_0003` supported 0.86 ### Context Scale - Total files: 215 - Important-file coverage: 40/215 - Evidence index entries: 79 - Role / Skill entries: 3 ### Handling Insufficient Evidence - **missing_evidence**: State that evidence is insufficient and ask the user for the target file, a README section, or after-install verification records; do not fill in facts. - **out_of_scope_request**: State that the task is beyond the current AI Context Pack's evidence scope and suggest the user check the Human Manual or verify after a real install. - **runtime_request**: Provide a pre-install checklist and command sources, but do not run commands for the user or claim they have been run. - **source_conflict**: Show the conflicting sources side by side, mark them as unverified, and do not force a single version. ## Prompt Recipes ### Fit assessment - Goal: Judge whether this project fits the user's current task. - Expected output: A fit conclusion, key reasons, evidence citations, what can be previewed before install, what must be verified after install, and a next-step recommendation. ```text Based on the AI Context Pack for total-recall, ask me 3 necessary questions first, then judge whether it fits my task. The answer must cover: who it fits, what it can do, what it cannot do, whether it is worth installing, and where the evidence comes from. Every project fact must cite evidence_refs, source_paths, or a claim_id. ``` ### Pre-install experience - Goal: Let the user feel the core workflow before installing, while avoiding packaging the preview as real capability or a marketing promise. - Expected output: An experience script with boundary labels, an after-install verification checklist, and a cautious recommendation; with no real-run promises or strong marketing language. ```text Treat total-recall as a pre-install experience asset, not an already-installed tool or a real runtime environment. Output exactly four parts: 1. Ask me 3 necessary questions first. 2. Give an "experience script": use the three labels [Previewable before install], [Must verify after install], and [Insufficient evidence] to show how it might guide the workflow. 3. Give an after-install verification checklist: list which capabilities can only be confirmed after a real install, real host loading, and a real project run. 4. Give a cautious recommendation: only "worth researching/trialing further", "add information before deciding", or "not recommended to continue"; do not endorse the project. Hard boundaries: - Do not claim you have installed, run, executed tests, modified files, or produced real results. - Do not write promise-like phrasing such as "auto-adapts", "guarantees passing", "perfect fit", or "strongly recommend installing". - If you describe how it works after install, you must use a conditional such as "if installed successfully and the host loads the Skill correctly, it might...". - The experience script may only be written as "example lines / hypothetical flow": use "might ask / might suggest / might show", not "has written, has generated, has passed, is running, is generating". - Prompt Preview does not hand out install commands; if the user is ready to trial, only prompt them to read Quick Start and the Risk Card first and to verify in an isolated environment. - Every project fact must come from a supported claim, evidence_refs, or source_paths; inferred/unverified items can only be risks or open questions. ``` ### Role / Skill selection - Goal: Pick the best-matching asset from the project's roles or Skills. - Expected output: A list of candidate roles or Skills, each with an applicable scenario, evidence paths, risk boundary, and whether after-install verification is needed. ```text Read role_skill_index and recommend 3-5 of the most relevant roles or Skills for my target task. For each recommendation, state the applicable scenario, likely output, risk boundary, and evidence_refs. ``` ### Risk pre-check - Goal: Identify environment, permission, rule-conflict, and quality risks before installing or adopting. - Expected output: A checklist of environment, permission, dependency, license, host-conflict, quality risk, and unknown items. ```text Based on risk_card, boundaries, and quick_start_candidates, give me a pre-install risk pre-check list. Do not run commands for me; only explain what I should check, why, and what impact a failure would have. ``` ### Host AI kickoff instruction - Goal: Turn the project context into a host AI instruction for the start of a conversation. - Expected output: A pre-work instruction with clear boundaries and clear evidence citations, suitable to copy to a host AI. ```text Based on the AI Context Pack for total-recall, generate a pre-work instruction I can paste to my host AI. This instruction must obey not_runtime=true and must not claim the project has been installed, run, or produced real results. ``` ## Role / Skill Index - Indexed 3 role / Skill / project-doc entries. - **total-recall:llm-setup** (skill): One-time setup for total-recall's product-owned ollama — installs the managed binary if needed and pulls default models. Triggered when SessionStart reports product ollama / models missing. Activation hint: When the user's task is highly relevant to the workflow described by “total-recall:llm-setup”, use it for a pre-install experience first, then decide whether to install. Evidence: `skills/llm-setup/SKILL.md` - **recall** (skill): Orient Claude on mining past Claude Code session logs via the total-recall plugin. Use when the user references prior work, asks "what did I do", asks about past decisions, mentions earlier sessions, wants to avoid repeating a corrected mistake, or asks why a previous choice was made. Surfaces user corrections, decisions, self-corrections, progress markers, and stable domain facts from ~/.claude/projects/ .jsonl acr… Activation hint: When the user's task is highly relevant to the workflow described by “recall”, use it for a pre-install experience first, then decide whether to install. Evidence: `skills/recall/SKILL.md` - **speak-like-operator** (skill): Match the operator's communication cadence lowercase, terse, no preambles, "we" framing, no emojis . Use when responding in chat. Don't mock their typos — match their directness. Activation hint: When the user's task is highly relevant to the workflow described by “speak-like-operator”, use it for a pre-install experience first, then decide whether to install. Evidence: `skills/speak-like-operator/SKILL.md` ## Evidence Index - Indexed 79 evidence entries. - **Installing total-recall in an agent CLI** (documentation): Installing total-recall in an agent CLI Evidence: `docs/install/README.md` - **Total Recall** (documentation): Memory & RAG for Claude Code and Grok — mines your own transcripts so new sessions already know decisions, bans, corrections, and goals. Evidence: `README.md` - **CLAUDE.md** (documentation): This file provides guidance to Claude Code claude.ai/code when working with code in this repository. Evidence: `CLAUDE.md` - **total-recall LLM + embed setup** (skill_instruction): Runs the operator-facing setup script that installs product ollama plugin data dir, daemon on :11435 and pulls: Evidence: `skills/llm-setup/SKILL.md` - **total-recall — Cross-session memory protocol** (skill_instruction): total-recall — Cross-session memory protocol Evidence: `skills/recall/SKILL.md` - **speak-like-operator — voice-matching protocol** (skill_instruction): speak-like-operator — voice-matching protocol Evidence: `skills/speak-like-operator/SKILL.md` - **Marketplace** (structured_config): { "$schema": "https://anthropic.com/claude-code/marketplace.schema.json", "name": "total-recall", "description": "Cross-session, cross-CLI memory for AI coding assistants. Mines transcripts from 10 supported CLI clients. Surfaces operator identity / decisions / bans / correc", "owner": { "name": "88plug", "email": "andrew@88plug.com" }, "plugins": { "name": "total-recall", "source": ".", "displayName": "Total Recall", "description": "Cross-session, cross-CLI memory for AI coding assistants. Mines transcripts from 10 supported CLI clients. Surfaces operator identity / decisions / bans / corrections / goals / voice via 26 MCP tools, 6 hooks SessionStart signpost + SessionStart compact-restore… Evidence: `.claude-plugin/marketplace.json` - **Plugin** (structured_config): { "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json", "name": "total-recall", "displayName": "Total Recall", "description": "Cross-session, cross-CLI memory for AI coding assistants. Mines transcripts from 10 supported CLI clients. Surfaces operator identity / decisions / bans / corrections / goals / voice via 26 MCP tools, 6 hooks SessionStart signpost + SessionStart compact-restore + UserPromptSubmit retrieval + Stop/PostCompact re-index + PreCompact continuity-seed , 15 slash commands, and 3 skills. Hybrid FTS5 + ollama dense recall qwen3-embedding:0.6b, default-on; TOTAL RECALL EMBED MODEL unset is correct . Product ollama auto-provisions refine qwen3.5:2b; TOTAL… Evidence: `.claude-plugin/plugin.json` - **Functional Source License, Version 1.1, ALv2 Future License** (source_file): Functional Source License, Version 1.1, ALv2 Future License Evidence: `LICENSE` - **.mcp.json notes** (documentation): Sibling file because JSON has no comments — anything explanatory lives here. Evidence: `.mcp.json.README.md` - **Recall Check Banned** (documentation): Call the check banned MCP tool with thing="$ARGUMENTS" . Evidence: `commands/recall-check-banned.md` - **Recall Corrections** (documentation): If $ARGUMENTS is non-empty, call recall corrections about with topic="$ARGUMENTS" and limit=10 . Otherwise call get recent corrections with the current working directory and limit=5 . Evidence: `commands/recall-corrections.md` - **Recall Cost** (documentation): Run bash "${CLAUDE PLUGIN ROOT}/scripts/recall-cli.sh" metrics cost --since ${ARGUMENTS:-30d} via Bash. Evidence: `commands/recall-cost.md` - **Recall Decisions** (documentation): If $ARGUMENTS is non-empty, call get decision for topic with topic="$ARGUMENTS" . Otherwise call list standing decisions with scope="global" . Evidence: `commands/recall-decisions.md` - **Recall Escalation** (documentation): Call the assess escalation risk MCP tool, passing the operator's last user message as last user and the draft assistant response if one exists in scratch as draft response . Evidence: `commands/recall-escalation.md` - **Recall Goal** (documentation): Call get active goal with cwd=$ARGUMENTS or the current working directory if $ARGUMENTS is empty to show the top-of-stack goal, then call list goals with the same cwd for the surrounding stack. Evidence: `commands/recall-goal.md` - **Recall Rebuild** (documentation): Summarize and confirm: - Drops DB at ${CLAUDE PLUGIN DATA}/total-recall/index.db . - Re-ingests every ~/.claude/projects/ / .jsonl . - Estimated 1-5 minutes for ~1GB on a dev laptop. Evidence: `commands/recall-rebuild.md` - **Architecture** (documentation): total-recall is a 4-layer pipeline. Each layer has one responsibility, depends only on the layer below it, and can be tested in isolation. The shape is deliberately boring: a walker feeds extractors, extractors feed an index, the index feeds delivery surfaces. Evidence: `docs/architecture.md` - **CI / CD architecture** (documentation): Runs on every push to main / master and on every pull request. Concurrent runs on the same ref are cancelled only the latest matters . Evidence: `docs/ci.md` - **Dense embeddings format v2 — product-owned ollama** (documentation): Dense embeddings format v2 — product-owned ollama Evidence: `docs/embeddings.md` - **Adversarial 10× eval** (documentation): Generated: 2026-07-18 01:29:34 -0400 Evidence: `docs/eval-adversarial-10x.md` - **LLM refine bakeoff total-recall chat model** (documentation): LLM refine bakeoff total-recall chat model Evidence: `docs/eval-llm-bakeoff.md` - **total-recall product model eval** (documentation): Generated: 2026-07-18 01:27:13 -0400 Evidence: `docs/eval-product-models.md` - **total-recall eval round 2 brand-new suites** (documentation): total-recall eval round 2 brand-new suites Evidence: `docs/eval-round2.md` - **total-recall eval round 3 — vectors live + 10x hard A/B** (documentation): total-recall eval round 3 — vectors live + 10x hard A/B Evidence: `docs/eval-round3.md` - **Scoreboard: product qwen stack vs legacy / FTS** (documentation): Scoreboard: product qwen stack vs legacy / FTS Evidence: `docs/eval-scoreboard.md` - **Total Recall** (documentation): Memory & RAG for Claude Code and Grok. Mines your own session transcripts so a new session already knows your decisions, corrections, bans, and goals. Evidence: `docs/index.md` - **Local-LLM refinement** (documentation): total-recall is 100% deterministic heuristics by default — zero model calls, zero network egress. The local-LLM refinement layer adds a second pass that improves machine-name extraction accuracy, vocabulary definitions, and per-project narrative summaries. Evidence: `docs/llm-refinement.md` - **total-recall on the 88plug marketplace** (documentation): total-recall on the 88plug marketplace Evidence: `docs/marketplace.md` - **Ollama GPU notes optional host tuning** (documentation): Ollama GPU notes optional host tuning Evidence: `docs/ollama-gpu.md` - **total-recall in Aider** (documentation): Status - MCP support: no Aider has no MCP client today - Hook support: no - Session storage: .aider.chat.history.md and .aider.input.history at each git-repo root - Adapter complexity: ~280 LOC Evidence: `docs/install/aider.md` - **total-recall in Claude Code** (documentation): Status - MCP support: yes - Hook support: yes SessionStart, UserPromptSubmit, Stop, PreCompact, PostCompact - Session storage: ~/.claude/projects/ / .jsonl - Adapter complexity: ~100 LOC thin wrapper over lib.jsonl walker Evidence: `docs/install/claude_code.md` - **total-recall in Cline** (documentation): Status - MCP support: yes ~/.cline/mcp.json or the VSCode UI under "Cline: MCP Servers" - Hook support: no - Session storage: Cline stores per-task conversations under its VSCode extension state; ingest support is planned no lib.sources.cline adapter ships yet at the time of writing . - Adapter complexity: pending Evidence: `docs/install/cline.md` - **total-recall in Codex CLI** (documentation): Status - MCP support: yes stdio, configured via ~/.codex/config.toml - Hook support: no - Session storage: ${CODEX HOME:-~/.codex}/sessions/ .jsonl - Adapter complexity: ~570 LOC replay state machine + token-block normalisation Evidence: `docs/install/codex.md` - **total-recall in Continue** (documentation): Status - MCP support: yes YAML config, ~/.continue/config.yaml - Hook support: no - Session storage: ~/.continue/sessions/ .json JSON-per-session - Adapter complexity: ~430 LOC Evidence: `docs/install/continue.md` - **total-recall in Cursor** (documentation): Status - MCP support: yes stdio, via ~/.cursor/mcp.json global or /.cursor/mcp.json per-project - Hook support: no - Session storage: ~/.cursor/projects/ /agent-transcripts/ .jsonl v1 / current CLI , state.vscdb v2 / planned - Adapter complexity: ~370 LOC JSONL only at v1 Evidence: `docs/install/cursor.md` - **total-recall in Gemini CLI** (documentation): Status - MCP support: yes stdio transport, via ~/.gemini/settings.json or gemini mcp add - Hook support: no - Session storage: ~/.gemini/tmp/ /logs.json replay-style - Adapter complexity: ~580 LOC replay reconstruction + tool-call translation Evidence: `docs/install/gemini_cli.md` - **total-recall in Goose** (documentation): Status - MCP support: yes via Claude Code or any MCP-capable host - Hook support: no - Session storage: ~/.local/share/goose/sessions/sessions.db SQLite - Adapter complexity: ~220 LOC Evidence: `docs/install/goose.md` - **total-recall in Grok** (documentation): Status - MCP support: yes via Claude Code or any MCP-capable host - Hook support: no - Session storage: ~/.grok/sessions/ / /chat history.jsonl - Adapter complexity: ~180 LOC Evidence: `docs/install/grok.md` - **total-recall in OpenCode** (documentation): Status - MCP support: yes local MCP type, stdio transport - Hook support: no OpenCode has no hook system at the time of writing - Session storage: ${OPENCODE DATA DIR:-~/.local/share/opencode}/opencode.db SQLite or .../storage/ legacy JSON - Adapter complexity: ~640 LOC two storage backends to translate Evidence: `docs/install/opencode.md` - **Init** (source_file): all = "EscalationAssessment", "assess escalation" Evidence: `detector/__init__.py` - **Init** (source_file): all = Evidence: `extractors/__init__.py` - **Base** (source_file): log = logging.getLogger name ⋮---- @runtime checkable class RecordLike Protocol ⋮---- type: str uuid: str parent uuid: str None session id: str cwd: str ts: datetime role: str None content kind: str None content: Any text: str None is meta: bool is compact summary: bool is sidechain: bool subtype: str None payload: dict None ⋮---- @runtime checkable class DagLike Protocol ⋮---- def get self, uuid: str - RecordLike None: ... def parent of self, uuid: str - RecordLike None: ... def next user turn self, uuid: str, within: int = 5 - RecordLike None: ... def prev assistant turn self, uuid: str - RecordLike None: ... ⋮---- @dataclass class Extraction ⋮---- kind: str content: str ⋮---- source uuid… Evidence: `extractors/base.py` - **Init** (source_file): all = Evidence: `index/__init__.py` - **Prefer longer tokens less stemming surprise with porter .** (source_file): all = ⋮---- def xdg data dir - Path ⋮---- def plugin data candidates - list Path ⋮---- root = Path "~/.claude/plugins/data" .expanduser ⋮---- found: list Path = ⋮---- nested = child / "total-recall" / "index.db" ⋮---- out: list Path = seen: set Path = set ⋮---- rp = p.resolve ⋮---- def list index candidates - list Path ⋮---- paths: list Path = ⋮---- p = d / "index.db" ⋮---- xdg = xdg data dir / "index.db" ⋮---- r = p.resolve ⋮---- r = p ⋮---- def resolve data dir - Path ⋮---- explicit dir = os.environ.get "TOTAL RECALL DB DIR" or "" .strip ⋮---- explicit file = os.environ.get "TOTAL RECALL DB" or "" .strip ⋮---- plugin data = os.environ.get "CLAUDE PLUGIN DATA" or "" .strip ⋮---- plugin dir… Evidence: `index/db.py` - **Init** (source_file): all = Evidence: `lib/__init__.py` - **Dag** (source_file): LinearizePolicy = Literal "deepest", "latest ts", "specified leaf" ⋮---- @dataclass class Dag ⋮---- nodes: dict str, Record = field default factory=dict children: dict str, list str = field default factory=dict roots: list str = field default factory=list leaves: list str = field default factory=list ⋮---- def eligible r: Record - bool ⋮---- def build dag records: list Record - Dag ⋮---- dag = Dag ⋮---- uuid = r.uuid ⋮---- def find branches dag: Dag - list tuple str, list str ⋮---- def walk up dag: Dag, leaf uuid: str - list Record ⋮---- path: list Record = cur: str None = leaf uuid seen: set str = set ⋮---- rec = dag.nodes.get cur ⋮---- cur = rec.parent uuid if rec.parent uuid in dag.nodes… Evidence: `lib/dag.py` - **Jsonl Walker** (source_file): DEFAULT PROJECTS ROOT = Path "~/.claude/projects" .expanduser ⋮---- def derive cwd from slug slug: str - str ⋮---- def derive slug from cwd cwd: str - str ⋮---- def session file filter p: Path - bool ⋮---- def session files for cwd projects root: Path, cwd: str - list Path ⋮---- slug = derive slug from cwd cwd proj dir = projects root / slug ⋮---- out = p for p in proj dir.iterdir if session file filter p ⋮---- cwd = derive cwd from slug proj.name ⋮---- def iter records path: Path, start offset: int = 0 - Iterator tuple int, Record ⋮---- path = Path path ⋮---- offset = start offset ⋮---- line = f.readline ⋮---- line len = len line ⋮---- stripped = line.strip ⋮---- obj = json.loads stripped.… Evidence: `lib/jsonl_walker.py` - **Schema** (source_file): ContentKind = Literal "string", "tool result", "text array", "image", "document", "empty" ⋮---- @dataclass class ToolUseRef ⋮---- id: str name: str input: dict str, Any ⋮---- @dataclass class Block ⋮---- type: str text: str None = None thinking: str None = None thinking signature: str None = None tool use: ToolUseRef None = None raw: dict str, Any = field default factory=dict ⋮---- @dataclass class ToolResult ⋮---- tool use id: str is error: bool content: str raw content: Any = None ⋮---- @dataclass class Record ⋮---- uuid: str None parent uuid: str None session id: str None ts: datetime None cwd: str None git branch: str None version: str None is sidechain: bool raw: dict str, Any byte off… Evidence: `lib/schema.py` - **Sidechain** (source_file): AGENT FILE RE = re.compile r"^agent- A-Za-z0-9 + \.jsonl$" ⋮---- def meta path for agent jsonl: Path - Path ⋮---- def read meta agent jsonl: Path - dict ⋮---- mp = meta path for agent jsonl ⋮---- obj = json.load f ⋮---- def subagent dir for session jsonl: Path - Path ⋮---- def subagent files for session jsonl path: Path - list tuple str, Path ⋮---- sub dir = subagent dir for Path jsonl path ⋮---- out: list tuple str, Path = ⋮---- m = AGENT FILE RE.match p.name ⋮---- workflows dir = sub dir / "workflows" ⋮---- def load subagent agent jsonl: Path - list Record ⋮---- def load subagent meta agent jsonl: Path - dict ⋮---- def parent tool use id for agent parent records: list Record , agent id: s… Evidence: `lib/sidechain.py` - **Init** (source_file): all = "mcp", "main" Evidence: `mcp_server/__init__.py` - **Dense embeds: product-owned ollama binary GitHub ollama/ollama latest 0.32.1 .** (source_file): build-system requires = "setuptools =77", "wheel" build-backend = "setuptools.build meta" Evidence: `pyproject.toml` - **Init** (source_file): version = "2.3.17" ⋮---- all = " version " Evidence: `total_recall/__init__.py` - **Cmd Llm** (source_file): @click.command "llm-model", help="Print the auto-selected local LLM model tag single line ." def llm model cmd - None ⋮---- tag = os.environ.get "TOTAL RECALL LLM MODEL" ⋮---- tag = autoselect model Evidence: `total_recall/cmd_llm.py` - **Also remove WAL/SHM siblings so we don't carry stale state.** (source_file): def text refine enabled env value: str None - bool ⋮---- def vec refine enabled env value: str None - bool ⋮---- def backfill vectors db path: str Path, verbose: bool - None ⋮---- status = ensure product ollama embed=True, chat=False, pull=True ⋮---- embedder = Embedder conn = connect db path ⋮---- report = backfill all conn, embedder=embedder ⋮---- db path = resolve db path ctx.obj.get "db path" verbose = bool ctx.obj.get "verbose" ⋮---- from index.db import apply schema, connect type: ignore import-not-found from index.ingest import ingest all type: ignore import-not-found ⋮---- Also remove WAL/SHM siblings so we don't carry stale state. ⋮---- p = Path str db path + suffix ⋮---- Recreate… Evidence: `total_recall/cmd_rebuild.py` - **Init** (source_file): all = ⋮---- def getattr name: str Evidence: `vec/__init__.py` - **Init** (source_file): all = Evidence: `extractors/llm/__init__.py` - **Init** (source_file): ClaudeCodeSource = None ⋮---- OpenCodeSource = None ⋮---- GeminiCliSource = None ⋮---- CodexSource = None ⋮---- CursorSource = None ⋮---- ContinueSource = None ⋮---- ClineSource = None ⋮---- AiderSource = None ⋮---- GooseSource = None ⋮---- GrokSource = None ⋮---- all = Evidence: `lib/sources/__init__.py` - **Base** (source_file): @dataclass class SessionFile ⋮---- source: str ⋮---- path: Path ⋮---- cwd: str None ⋮---- session id: str ⋮---- started at: float None ⋮---- last modified: float ⋮---- extra: dict = field default factory=dict ⋮---- class SessionSource ABC ⋮---- name: str ⋮---- @abstractmethod def is available self - bool ⋮---- @abstractmethod def discover sessions self - Iterator SessionFile ⋮---- SOURCES: list type SessionSource = ⋮---- def all sources - list SessionSource ⋮---- def source by name name: str - SessionSource None Evidence: `lib/sources/base.py` - **toolResponse should not appear on assistant turns, but keep** (source_file): log = logging.getLogger name ⋮---- def parse unix seconds raw: Any - datetime None ⋮---- def parse sql timestamp raw: Any - float None ⋮---- def decode value: Any - Any ⋮---- raw = value if isinstance value, str else value.decode "utf-8", errors="replace" ⋮---- def flatten tool result content value: Any - tuple str, Any ⋮---- parts: list str = ⋮---- t = b.get "text" ⋮---- sc = value.get "structuredContent" ⋮---- out = sc.get "stdout" ⋮---- def assistant blocks content: Any - list Block ⋮---- """Normalize a Goose assistant content json array into Blocks.""" ⋮---- out: list Block = ⋮---- bt = b.get "type", "?" ⋮---- call = b.get "toolCall" or {} val = call.get "value" if isinstance call, dict… Evidence: `lib/sources/goose.py` - The remaining 19 evidence entries are in `AI_CONTEXT_PACK.json` or `EVIDENCE_INDEX.json`. ## Rules the Host AI Must Follow - **Treat this asset as pre-work context, not a runtime environment.**: The AI Context Pack contains only an evidence-backed understanding of the project, not the project's executable state. Evidence: `docs/install/README.md`, `README.md`, `CLAUDE.md` - **When answering the user, distinguish what can be previewed from what can only be verified after install.**: The consumer value of the pre-install experience comes from reducing bad installs and misjudgments, not from pretending to be a real run. Evidence: `docs/install/README.md`, `README.md`, `CLAUDE.md` ## Questions the User Should Answer First - Which host AI or local environment do you plan to use it in? - Do you just want to experience the workflow first, or are you ready to actually install? - What matters most to you: install cost, output quality, or conflicts with your existing rules? ## Acceptance Checks - Every capability claim can be traced back to a file path in evidence_refs. - AI_CONTEXT_PACK.md does not package previews as a real run. - The user can understand who it fits, what it can do, how to start, and the risk boundaries within 3 minutes. --- ## Doramagic Context Augmentation The following sections strengthen the repository context for a host AI. Human Manual data is a reading route, and pitfall notes become operating constraints. ## Human Manual Outline Usage rule: this is only a reading route and salience signal, not factual authority. Concrete claims must still return to repo evidence or Claim Graph. Host AI hard rules: - Do not treat page titles, section order, summaries, or importance values as factual project evidence. - When explaining the Human Manual outline, state that it is only a reading route or salience signal. - Capability, installation, compatibility, runtime state, and risk claims must cite repo evidence, source paths, or Claim Graph. - **Overview**: importance `high` - source_paths: README.md, pyproject.toml - **Lib**: importance `high` - source_paths: hooks/lib/build_packet.py, hooks/lib/common.sh, hooks/lib/compact_restore.py, hooks/lib/decide_and_format.py, hooks/lib/query.py - **Lib**: importance `high` - source_paths: lib/__init__.py, lib/dag.py, lib/jsonl_walker.py, lib/schema.py, lib/sidechain.py - **Commands**: importance `high` - source_paths: commands/recall-check-banned.md, commands/recall-corrections.md, commands/recall-cost.md, commands/recall-decisions.md, commands/recall-escalation.md ## Repo Inspection Evidence - repo_clone_verified: true - repo_inspection_verified: true - repo_commit: `e5cb2dc23cb9e45416f1f2fa6274ecf83463d613` - inspected_files: `README.md`, `pyproject.toml`, `uv.lock`, `docs/architecture.md`, `docs/ci.md`, `docs/embeddings.md`, `docs/eval-adversarial-10x.md`, `docs/eval-llm-bakeoff.md`, `docs/eval-product-models.md`, `docs/eval-round2.md`, `docs/eval-round3.md`, `docs/eval-scoreboard.md`, `docs/index.md`, `docs/install/README.md`, `docs/install/aider.md`, `docs/install/claude_code.md`, `docs/install/cline.md`, `docs/install/codex.md`, `docs/install/continue.md`, `docs/install/cursor.md` Host AI hard rules: - Without repo_clone_verified=true, do not claim that the source code has been read. - Without repo_inspection_verified=true, do not write README, docs, or package-file conclusions as facts. - Without quick_start_verified=true, do not claim that the Quick Start path has run successfully. ## Doramagic Pitfall Constraints These rules come from Doramagic discovery, validation, or compilation findings. The host AI must treat them as operating constraints, not background notes. ### Constraint 1: Capability evidence risk requires verification - Trigger: README/documentation is current enough for a first validation pass. - Host AI rule: Reproduce the official install and quickstart path in an isolated environment. - Why it matters: May increase setup, validation, or first-run risk for the user. - Evidence: capability.assumptions | https://github.com/88plug/total-recall - Hard boundary: Do not present this pitfall as solved, verified, or ignorable unless later evidence explicitly closes it. ### Constraint 2: Security or permission risk requires verification - Trigger: no_demo - Host AI rule: Reproduce the official install and quickstart path in an isolated environment. - Why it matters: May increase setup, validation, or first-run risk for the user. - Evidence: downstream_validation.risk_items | https://github.com/88plug/total-recall - Hard boundary: Do not present this pitfall as solved, verified, or ignorable unless later evidence explicitly closes it. ### Constraint 3: Security or permission risk requires verification - Trigger: no_demo - Host AI rule: Reproduce the official install and quickstart path in an isolated environment. - Why it matters: May increase setup, validation, or first-run risk for the user. - Evidence: risks.scoring_risks | https://github.com/88plug/total-recall - Hard boundary: Do not present this pitfall as solved, verified, or ignorable unless later evidence explicitly closes it. ### Constraint 4: Maintenance risk requires verification - Trigger: issue_or_pr_quality=unknown。 - Host AI rule: Reproduce the official install and quickstart path in an isolated environment. - Why it matters: May increase setup, validation, or first-run risk for the user. - Evidence: evidence.maintainer_signals | https://github.com/88plug/total-recall - Hard boundary: Do not present this pitfall as solved, verified, or ignorable unless later evidence explicitly closes it. ### Constraint 5: Maintenance risk requires verification - Trigger: release_recency=unknown。 - Host AI rule: Reproduce the official install and quickstart path in an isolated environment. - Why it matters: May increase setup, validation, or first-run risk for the user. - Evidence: evidence.maintainer_signals | https://github.com/88plug/total-recall - Hard boundary: Do not present this pitfall as solved, verified, or ignorable unless later evidence explicitly closes it.