Tages

Store a memory about this codebase — conventions, decisions, architecture, patterns, or lessons learned

Search codebase memories by fuzzy query — finds conventions, decisions, patterns, and lessons. Results are numbered passages [1..n] with source and date. Before answering, briefly note which numbered passage(s) support your answer, then cite them.

Delete a memory by its key

List all coding conventions for this project

List all architecture notes and module boundaries

List the decision log — why things were built the way they are

Get all memories related to a specific file path

Check for stale memories that may be outdated or no longer relevant

Detect potential conflicts between memories — overlapping or contradictory entries

Show memory usage statistics — counts by type, recall hit rate, agent sessions, most/least accessed

Report what you are doing or learning — Tages silently extracts memories from your observations. Call this naturally as you work, no need to format.

End the current session with a summary of what was built, decided, or learned — auto-extracts memories from the summary

Verify a pending auto-extracted memory — promotes it to live so it appears in recall results

List auto-extracted memories that need verification before they appear in recall

Detailed memory statistics — counts by type and status, average confidence, top 5 agents, and total count

Get the version history of a memory — shows past values, confidence changes, and who changed it. Optionally revert to a previous version.

Search memories filtered by execution context — current files, agent name, or phase. More precise than recall for context-aware retrieval.

Resolve a detected memory conflict using keep_newer, keep_older, or merge strategies

List unresolved memory conflicts for this project

Get suggestions for memories you should store, based on queries that returned no results

Import memories from a JSON array or markdown file content. Handles duplicates with skip/overwrite/merge strategies.

Build a relationship graph from crossSystemRefs and output as a Mermaid diagram

Fork current memories into a session branch for experimentation. Branch writes won't affect main until merged.

Merge a session branch back into main memory. Detects conflicts between branch and main changes.

List memories in a session branch

Detect near-duplicate memories using Jaccard similarity on key+value tokens

Merge two duplicate memories into one, preserving all metadata

Analyze the downstream impact of a memory — how many others depend on it

Get project-wide risk ranking — which memories are most dangerous to change

Analyze the memory dependency graph — orphans, critical paths, impact scores

Check a memory against all stored conventions — detect conflicts or duplicates before storing

Get per-agent convention compliance stats and recent violations

Score a specific memory on completeness, freshness, consistency, and usefulness (0-100)

Get overall project memory health score with dimension breakdown and improvement suggestions

List available memory templates (api-endpoint, react-component, database-migration, test-suite, cli-command)

Find templates matching given file paths — prompts agent to fill required fields

Create a memory from a filled template with automatic key generation

Archive a memory to cold storage — removes it from recall but preserves it for restoration

Restore an archived memory back to live status

List archived memories with their archive reasons and dates

Get archive statistics — total archived, restored, expired

Scan for stale low-quality memories and auto-archive them

Promote a project memory to the org-wide federated library

Import a federated memory from the shared library into this project

List all memories in the federated library

Show which federated memories have local project overrides

Replay a session timeline showing all tool calls with metrics

Get agent effectiveness metrics — recall hit rate, memory creation quality, convention compliance

Detect performance trends across sessions — improvements and regressions

Before starting a task, get a list of gotchas: anti-patterns to avoid, conventions to follow, and lessons learned from past experience

Find memories related to specific file paths — matches by exact path, directory prefix, or reverse prefix. Prioritises anti-pattern and convention memories.

Generate a token-budgeted project brief for system prompt injection. Returns gotchas, conventions, architecture, and decisions — prioritized by importance. Use this ONCE at session start instead of calling recall/conventions/architecture individually.

Parse a CLAUDE.md file and auto-create memories from its sections. Conventions, architecture notes, decisions, and anti-patterns become typed memories.

Audit project memory coverage and quality: type distribution, brief-critical coverage score, imperative phrasing ratio, and actionable suggestions.

Rewrite a memory value into imperative form (ALWAYS/NEVER/MUST/DO NOT) using Claude Haiku. Returns a before/after preview unless confirmed=true.

End-of-session tool: extract memories from a session summary, optionally regenerate the project brief. Combines session_end + brief regeneration in one call.