agentleFS
huguryildizhuguryildiz/ALETHEIA/skills/core/decision-log/SKILL.md

decision-log

Skill0 starsChanged 4 months agoMITView on GitHub

Use when a project decision is made or changed — a modeling assumption, a default value, a method choice, a scope cut, a rejected alternative — and it must be recorded as a numbered ADR-lite entry, or when an existing decision is superseded, or when the user asks "why did we choose X" / "is there a decision for this". Trigger phrases: "record this decision", "write an ADR", "add a D-number", "supersede D12", "why is this the default", "log the rationale".

46 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use product-brief, pre-mortem and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • erik-kroonSkill

    product-brief

    Turn conversation context, codebase findings, and product intent into a PRD-style brief without assuming a specific issue tracker. Use when the user wants a PRD, product brief, feature spec, or implementation-ready product artifact.

    Also in 3 projects that use this

  • Ingramradical235Skill

    pre-mortem

    Pre-mortem analysis, scenario planning, future state narratives, and contingency frameworks. Use before finalizing any Strategy Kernel, when planning major initiatives, or when preparing Plan B triggers.

    Also in 3 projects that use this

  • MicrosoftSkill

    retrospective

    Review how work went and improve the next round. USE WHEN asked for a retro, delays, or process lessons. DO NOT USE for code review or transcript replay.

    Also in 4 projects that use this

  • OpenAISkill

    meeting-prep

    Use when creating Public Equity Investing meeting prep briefs. Do not use for private diligence, IB, FP&A, legal, or scheduling-only tasks.

    Also in 3 projects that use this

  • Microsoft AzureSkill

    test

    Also in 3 projects that use this

  • HashiCorpSkill

    changelog

    Add a `.changelog/<PR_NUMBER>.txt` entry for a GitHub Pull Request, commit, and push.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 39 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Keep one append-mostly file — {{decision_log}} — holding the project's numbered decisions with status and rationale.

    line 10
  2. Very specific:

    Run evidence and numbers — those live in the build log and {{evidence_dir}}.

    line 30
  3. Very specific:

    Check for a predecessor: If this decision changes an existing one, write the new entry, then edit only the old entry's Status: line to …

    line 60

“Common” is measured: kept by at least 6 in 10 of the 39 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

decision-log

Keep one append-mostly file — {{decision_log}} — holding the project's numbered decisions with status and rationale. It is the project's authority layer: when code, docs, and notebooks disagree, the decision log arbitrates (see layer-sync). Entries are conclusions, not discussions; superseded entries stay in place with a changed status so the derivation chain of the eventual paper stays traceable.

When to use

  • An assumption, default, threshold, data source, or method is chosen — especially anything another layer (spec, code, notebook, paper) will depend on.
  • A previously accepted decision is replaced, relaxed, or reversed.
  • An alternative was seriously evaluated and rejected — negative decisions are entries too; they stop the team from re-litigating.
  • A decision is consciously deferred past the current publication target.

When NOT to use

  • Implementation details fully visible in the code and its tests — the diff is the record.
  • Transient exploration ("let's see this number") — that is a notebook note (project-layout), promoted to a decision only if it changes a default.
  • Run evidence and numbers — those live in the build log and {{evidence_dir}}; the decision entry links to them.
  • Rewriting history — never edit an old entry's substance; supersede it.

Entry format

Newest entry at the top, directly under the conventions header. Undated founding conventions live at the bottom.

## D<NN> — <one-line imperative title>

**Status**: accepted | superseded (→ D<MM>) | rejected | deferred
**Date**: YYYY-MM-DD

**Context**: why the question arose — 2–5 sentences, with links to the triggering
artifact (roadmap item, audit finding, notebook, failing gate).

**Decision**: the resolution, as numbered items when it has parts. State values
explicitly ("threshold = 0.85"), name the alternative that lost and in one line why.

**Impact**: which layers must change (spec §, module, notebook, status board), whether any
protected default changed (see canonical-params), and what evidence backs the choice
(link to {{evidence_dir}} run dirs or build-log entries).

Runbook

  1. Find the next free number: grep -oE '^## D[0-9]+' {{decision_log}} | sort -V | tail -1 and increment. Never reserve numbers in advance; the number is assigned at write time.
  2. Check for a predecessor. If this decision changes an existing one, write the new entry, then edit only the old entry's Status: line to superseded (→ D<new>).
  3. Write the entry at the top in the format above. Conclusion language, not deliberation transcripts; link the background instead of pasting it.
  4. Propagate or schedule. Either update the dependent layers in the same change, or state in Impact what remains — the next layer-sync audit will hold you to it.

Rules

  1. One decision per entry. Two questions resolved together are two entries that cross-reference each other.
  2. Statuses are exactly: accepted, superseded, rejected, deferred. A superseded entry keeps its body verbatim; only the status line changes.
  3. Every numeric default cites a justification — a measurement, a reference, or an explicit "design operating point, to be calibrated" honesty label. Unjustified constants are how silent errors enter papers.
  4. The Impact section is a propagation contract. Layers it names are what layer-sync audits; omitting a layer there does not exempt it, it just makes the drift harder to find.
  5. Cite decisions by number everywhere else (docs, notebooks, commit messages) — "per D14" — and gloss the meaning on first use in any document ("the count-matching decision (D14)"), so readers never need a lookup table.
  6. Findings can be decisions. A negative or infeasibility result that fixes the project's direction deserves an entry with the evidence linked — do not soften it.

Configuration

  • {{decision_log}} — the file (default docs/decisions.md).
  • {{evidence_dir}} — where linked run evidence lives.

Provenance & maintenance

Generalized from a 50+-entry ADR-lite chain in a working computational-research repository (newest-on-top ordering, four-status lifecycle, supersede-don't-edit, Impact-as-propagation- contract); see the pack's examples/ directory for the worked exemplar mapping. The format block is the exemplar's structure with domain content removed; the four status values and the next-free-number rule are verified practice.

Re-verify in your project:

  • grep -oE '^## D[0-9]+' {{decision_log}} | sort -V | tail -1 — highest number = next − 1; duplicates mean a numbering collision to fix.
  • grep -cE '\*\*Status\*\*: (accepted|superseded|rejected|deferred)' {{decision_log}} — should equal the entry count.
  • head -20 {{decision_log}} — the newest entry must be at the top.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.