accenture/mercury/archive-fact

Claude Code and agent skill by accenture/mercury

Trust tier 2
  • community
LLM review pending
Source check pending

Checking that the SKILL.md file is still in the repository.

Safely move a decayed fact or completed thread from continuity.md into the archive — the mechanical half of a memory review, done deterministically. Use during the REVIEW.md ritual (step 4) or whenever a specific fact must be archived, instead of hand-editing continuity.md (which has truncated the file twice). The agent decides what to archive; this performs the move safely.

Install

npx vskill@latest install accenture/mercury/archive-fact

SKILL.md

⚠️ Tool-managed skill provided by agent-memory (provenance: agent-memory-builtin). Don't edit it
in place — fork under a new name, or upstream a fix to the agent-memory project (see SKILLS.md).

This skill is the safe archive-move. It performs REVIEW.md step 4 — moving a faded/superseded fact's block out of the live layer — continuity.md, or a thread's own memory/open-threads/thread-<id>.md file (v4.39.0; the sweep moves the block to the quarter archive and deletes the file) — as a runnable script, so the move can't be botched. Don't hand-edit `continuity.md` to do this (the read-modify-write open(f,"w").write(open(f).read()+…) truncates the file before the read — it has wiped a version.md stamp and then this repo's archive, 50 facts → 6, once each). This script reads the whole file into memory first and writes once, so truncation is structurally impossible.

What it does NOT do

It does not decide what to archive — that judgment (which facts have faded, contradiction scans, supersession) stays with the agent and the REVIEW.md ritual (never-pick-a-winner). This is the mechanical executor only: you name the id(s); it moves them. Same split as memory-lint (the script counts; the agent judges).

How to run

From the repo root, use whichever runtime the machine has (output is byte-identical):

sh
python3 agent-skills/archive-fact/scripts/archive-fact.py --id <fact-id> [--id <fact-id> …]
# or
node agent-skills/archive-fact/scripts/archive-fact.mjs --id <fact-id> [--id <fact-id> …]

Flags:

  • --id ID — the fact/thread id to archive (repeatable; the move is all-or-nothing — if any id is

missing it refuses and changes nothing).

  • --reason — faded (default) or e.g. "superseded by <new-id>".
  • --quarter YYYY-Qn — target memory/archive/<quarter>.md (default: derived from today, UTC).
  • --note TEXT — the INDEX.md one-liner (single --id only; otherwise derived from the block).
  • --dry-run — print what would move; change nothing. Preview first when unsure.

It (1) extracts each id's block (the column-0 - bullet down through its <!-- id: … --> footer, whole), (2) appends it to the quarter archive under a dated heading and adds an INDEX.md line (append mode — never truncates), and (3) rewrites continuity.md without those blocks (read-into-variable, single write).

Always run `memory-lint` afterward (it confirms no id ended up in both places and no link dangles). The helper is safe by construction; the lint is the deterministic proof.

Guards (it refuses, exit 1)

  • an id with no footer in continuity.md or memory/open-threads/ (typo / already moved);
  • an id already present in the archive (no double-archive);
  • a move that would leave continuity.md empty.

Notes

  • No runtime? Don't hand-roll the truncating write. Fall back to the manual safe method REVIEW.md

documents (append-mode / read-into-variable), then run memory-lint. (Same stance as memory-lint: install Python or Node rather than improvise the dangerous step.)

  • Two runtimes at output parity, with mirror tests (test_archive_fact.py / .mjs) — the deterministic

guarantee shouldn't depend on which runtime the machine happens to have.

  • Superseded facts: pass --reason "superseded by <new-id>"; mark the supersession links in the footers

(DECAY.md §9) before or after — the helper moves the block as-is.

Security

  • Security scanPassed, 100 of 100Pattern checks on the skill's files.
  • Model reviewNot runA model review is scheduled for the next rescan.

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
91
Repository forks
40
7-day trend score
23

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/accenture/mercury/archive-fact/badge)](https://verified-skill.com/skills/accenture/mercury/archive-fact)

Something wrong with this skill or its page? Tell us.