attic-stash
Skill5 starsChanged 2 months agoMITView on GitHub
Stash the current finding, result, plan or decision into .attic/ and reply with its handle. Use when the user says "stash this", "put this in the attic", "remember this", "save this for later", or when a finding is worth keeping and attic mode is active.
In this repository12 other files it gives its agents.
- attic-doctorSkill
- attic-failSkill
- attic-gitSkill
- attic-helpSkill
- attic-indexSkill
- attic-pinSkill
- attic-pruneSkill
- attic-recallSkill
- attic-reviewSkill
- atticSkill
- attic-statsSkill
- attic-sweepSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 20
If nothing obvious exists, ask in one line what to stash.
- Distinctive: line 27
Write the body to a temp file, then call the script.
- Very specific: line 43
If the script cannot run at all (no Node, or the call is denied), fall back to writing the files yourself following …
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Stash into the attic
Argument: $ARGUMENTS (a title, a slug, or empty).
- Decide what to stash: the most recent finding, result, plan or decision in
this conversation. If the argument names something specific, stash that.
If nothing obvious exists, ask in one line what to stash.
Unsure whether it is worth keeping? Read
${CLAUDE_SKILL_DIR}/../attic/references/what-to-stash.md. - Pick a
kind(finding,decision,plan,output,note) and a specific kebab-case slug that will still make sense in a month. - Record the evidence (see Provenance below): which files the conclusion came from, which commands produced it, and how far you actually checked it.
- Write the body to a temp file, then call the script. It handles the folder, frontmatter, index line, atomic write, and secret scan:
node "${CLAUDE_SKILL_DIR}/../attic/scripts/attic.js" stash \
--slug <slug> --kind <kind> --title "<title>" \
--hook "<one line under 100 chars>" [--tags a,b] \
[--type <type>] [--confidence verified|unverified] \
[--files "src/a.ts,src/b.ts"] [--commands "rg foo src"] \
[--decision-why "<why>"] --body-file <tmpfile>An existing slug appends a dated update instead of overwriting.
- Exit code 2 means a credential was detected and nothing was written.
Redact the value, keep the location, retry. Never pass
--force. - If the script cannot run at all (no Node, or the call is denied), fall
back to writing the files yourself following
${CLAUDE_SKILL_DIR}/../attic/references/item-format.md, and say in one line that the secret scan and atomic write were skipped. Never silently substitute hand-written files for the script. - Reply with exactly:
Stashed `attic:<slug>` · <hook>.
Provenance
These flags are what make a finding re-checkable later. They are optional and the stash works without them, but a finding with no evidence cannot be told apart from a guess six sessions from now.
| Flag | Pass when |
--files | You read those files to reach the conclusion. Repo-relative. This is what freshness watches: if none are named, nothing can be checked. |
--commands | A command produced the evidence (rg ..., npm test). Lines containing a credential are dropped automatically. |
--confidence verified | Only when you actually checked the claim against the code in this session. |
--type | The item is a failed-approach or a workaround. Otherwise it follows kind. |
--confidence defaults to unverified, and that default is usually right.
Saving a finding is not the same as confirming it. Claim verified only for
something you read and checked; a plausible inference, a remembered detail, or
anything a user told you is unverified. Over-claiming here is worse than
saying nothing, because it disables the warning a later session would get.
Naming files you did not actually look at is the same error in a different place: it makes freshness watch the wrong thing.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

