Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Decision Logger · alirezarezvani | DecimalAI

alirezarezvani/decision-logger

SkillScore77Maintained

Two-layer memory architecture for board meeting decisions. Manages raw transcripts (Layer 1) and approved decisions (Layer 2). Use when logging decisions after a board meeting, reviewing past decisions with /cs:decisions, or checking overdue action items with /cs:review. Invoked automatically by the board-meeting skill after Phase 5 founder approval.

agent-toolingby alirezarezvani

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Knowledge Management pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/alirezarezvani-decision-logger/SKILL.md

decimalai skills pull alirezarezvani-decision-logger --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/alirezarezvani-decision-logger/SKILL.md https://app.decimal.ai/s/alirezarezvani-decision-logger/SKILL.md

Raw file: /s/alirezarezvani-decision-logger/SKILL.md ↗ · machine-readable summary at /s/alirezarezvani-decision-logger.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 25d ago

—+30% tokens0% turns77% of whole cases pass with skill (17/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-04	✗→✓	▲ Improved	19%	0%
case-05	✗→✓	▲ Improved	5%	0%
case-06	✗→✓	▲ Improved	7%	0%
case-07	✗→✓	▲ Improved	5%	0%
case-08	✗→✓	▲ Improved	24%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

alirezarezvani / claude-skills

.gemini/skills/decision-logger/SKILL.md

View on GitHub ↗

License MITImported 8/3/2026Version v1

Skill contents

SKILL.md

Decision Logger

Two-layer memory system. Layer 1 stores everything. Layer 2 stores only what the founder approved. Future meetings read Layer 2 only — this prevents hallucinated consensus from past debates bleeding into new deliberations.

Keywords

decision log, memory, approved decisions, action items, board minutes, /cs:decisions, /cs:review, conflict detection, DO_NOT_RESURFACE

Quick Start

bash
python scripts/decision_tracker.py --demo             # See sample output
python scripts/decision_tracker.py --summary          # Overview + overdue
python scripts/decision_tracker.py --overdue          # Past-deadline actions
python scripts/decision_tracker.py --conflicts        # Contradiction detection
python scripts/decision_tracker.py --owner "CTO"      # Filter by owner
python scripts/decision_tracker.py --search "pricing" # Search decisions

Commands

| Command | Effect | |---------|--------| | /cs:decisions | Last 10 approved decisions | | /cs:decisions --all | Full history | | /cs:decisions --owner CMO | Filter by owner | | /cs:decisions --topic pricing | Search by keyword | | /cs:review | Action items due within 7 days | | /cs:review --overdue | Items past deadline |

Two-Layer Architecture

Storage follows the canonical two-layer decision memory (see ../agent-protocol/SKILL.md → "Decision Memory (Canonical Layout)") — the same layout /cs:decide writes.

Layer 1 — Raw Transcripts

Location: ~/.claude/decisions/raw/YYYY-MM-DD-<slug>.md

Full Phase 2 agent contributions, Phase 3 critique, Phase 4 synthesis
All debates, including rejected arguments
NEVER auto-loaded. Only on explicit founder request.
Archive after 90 days → ~/.claude/decisions/raw/archive/YYYY/

Layer 2 — Approved Decisions

Location: ~/.claude/decisions/approved/ — one record per decision (YYYY-MM-DD-<slug>.md) plus the append-only index decisions.md

ONLY founder-approved decisions, action items, user corrections
Loaded automatically in Phase 1 of every board meeting
Append-only. Decisions are never deleted — only superseded.
Managed by Chief of Staff after Phase 5. Never written by agents directly.

Migration: a legacy memory/board-meetings/ folder may exist from earlier versions; read it for history but write all new entries to ~/.claude/decisions/.

Decision Entry Format

markdown
## [YYYY-MM-DD] — [AGENDA ITEM TITLE]

**Decision:** [One clear statement of what was decided.]
**Owner:** [One person or role — accountable for execution.]
**Deadline:** [YYYY-MM-DD]
**Review:** [YYYY-MM-DD]
**Rationale:** [Why this over alternatives. 1-2 sentences.]

**User Override:** [If founder changed agent recommendation — what and why. Blank if not applicable.]

**Rejected:**
- [Proposal] — [reason] [DO_NOT_RESURFACE]

**Action Items:**
- [ ] [Action] — Owner: [name] — Due: [YYYY-MM-DD] — Review: [YYYY-MM-DD]

**Supersedes:** [DATE of previous decision on same topic, if any]
**Superseded by:** [Filled in retroactively if overridden later]
**Raw transcript:** ~/.claude/decisions/raw/[DATE]-<slug>.md

Conflict Detection

Before logging, Chief of Staff checks for:

DO_NOT_RESURFACE violations — new decision matches a rejected proposal
Topic contradictions — two active decisions on same topic with different conclusions
Owner conflicts — same action assigned to different people in different decisions

When a conflict is found:

⚠️ DECISION CONFLICT
New: [text]
Conflicts with: [DATE] — [existing text]

Options: (1) Supersede old  (2) Merge  (3) Defer to founder

DO_NOT_RESURFACE enforcement:

🚫 BLOCKED: "[Proposal]" was rejected on [DATE]. Reason: [reason].
To reopen: founder must explicitly say "reopen [topic] from [DATE]".

Logging Workflow (Post Phase 5)

Founder approves synthesis
Write Layer 1 raw transcript → ~/.claude/decisions/raw/YYYY-MM-DD-<slug>.md
Check conflicts against ~/.claude/decisions/approved/decisions.md
Surface conflicts → wait for founder resolution
Write the approved record to ~/.claude/decisions/approved/YYYY-MM-DD-<slug>.md and append to the index decisions.md
Confirm: decisions logged, actions tracked, DO_NOT_RESURFACE flags added

Marking Actions Complete

markdown
- [x] [Action] — Owner: [name] — Completed: [DATE] — Result: [one sentence]

Never delete completed items. The history is the record.

File Structure

~/.claude/decisions/
├── raw/YYYY-MM-DD-<slug>.md        # Layer 1: full transcript per meeting
├── raw/archive/YYYY/               # Raw files after 90 days
├── approved/YYYY-MM-DD-<slug>.md   # Layer 2: one record per approved decision
└── approved/decisions.md           # Layer 2 index: append-only, founder-approved

References

templates/decision-entry.md — single entry template with field rules
scripts/decision_tracker.py — CLI parser, overdue tracker, conflict detector

Want to see if alirezarezvani/decision-logger would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/11/2026, 09:54 UTC · 22 cases

Measured on 20 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+59%

77% vs 18%

Agent turns

1.0 vs 1.0

Tokens

+30%

2,218.2 vs 1,710.25

fail → pass · 14pass → pass · 3fail → fail · 4pass → fail · 1

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-21 We need to export our meeting notes into a formatted HTML styled template so our corporate secretary can print them to PDF for legal archiving. Can you generate the HTML layout?	pass→pass	23,824	25,210	+6%	1	1	0%	5,256	6,849	+30%	0	0	—
case-01 Can you scan our recorded meeting memory and pull up a report of all action items that have missed their target deadlines, along with any tasks coming due this week?	fail→fail	4,526	9,411	+108%	1	1	0%	763	2,740	+259%	0	0	—
case-02 We just wrapped up our executive meeting and agreed to reallocate marketing budget to enterprise sales. Please record this approved consensus into our official decision system—include the core rationale, assign the VP of Sales as owner, list the rejected outbound proposal, and flag any conflicts with existing policy before finalizing the entry.	fail→fail	7,360	5,080	-31%	1	1	0%	1,254	1,706	+36%	0	0	—
case-03 I need to review our past approved board resolutions regarding our subscription tier pricing. Could you search our formal decision history for pricing records and list the active decisions along with who owns them?	fail→fail	3,015	5,831	+93%	1	1	0%	426	1,654	+288%	0	0	—
case-04 We want to store the complete unedited meeting debate from today's session—including all rejected agent arguments and counter-synthesis—into our raw memory archive. Where should this file be saved on disk, and what is the standard retention policy before moving it to historical archive folders?	fail→pass	9,955	2,584	-74%	1	1	0%	1,567	1,869	+19%	0	0	—
case-05 When initiating Phase 1 of a new board deliberation, should the agent system automatically read all raw debate transcripts from previous meetings to build context, or should it only read founder-approved decisions?	fail→pass	13,012	4,061	-69%	1	1	0%	2,012	2,106	+5%	0	0	—
case-06 During our last strategic review, the team rejected a proposal to switch from AWS to GCP due to vendor locking risks. How should this rejected item be flagged in the decision record so agents do not re-propose it in future meetings?	fail→pass	12,232	4,069	-67%	1	1	0%	1,973	2,116	+7%	0	0	—
case-07 An agent attempts to bring back an outbound sales campaign idea that was previously marked with a DO_NOT_RESURFACE tag. What exact explicit human authority or phrasing is required to bypass this restriction?	fail→pass	10,358	13,320	+29%	1	1	0%	1,629	1,709	+5%	0	0	—
case-08 When logging a new decision regarding server architecture, the system discovers a direct contradiction with an active 2024 resolution. What specific set of options must be presented to resolve the contradiction?	fail→pass	8,246	2,183	-74%	1	1	0%	1,393	1,733	+24%	0	0	—
case-09 We finished implementing the single sign-on task that was assigned in last month's security review. How should we update the action item in our approved decision ledger?	fail→pass	10,560	4,420	-58%	1	1	0%	1,692	2,151	+27%	0	0	—
case-10 I want to run a quick terminal script command to see all action items across all projects that have passed their deadline without deleting or altering any files. What CLI command and flag should I invoke?	fail→pass	9,335	1,805	-81%	1	1	0%	1,556	1,661	+7%	0	0	—
case-11 I need to inspect all decisions and action items assigned specifically to the CTO using our python command line tool. What exact arguments should be passed?	fail→pass	5,130	1,869	-64%	1	1	0%	842	1,659	+97%	0	0	—
case-12 What quick slash command outputs the 10 most recently approved decisions from our memory index?	fail→pass	4,934	1,495	-70%	1	1	0%	814	1,621	+99%	0	0	—
case-13 How can a user print the entire historical index of approved decisions using slash commands instead of limiting output to the default top 10?	fail→pass	10,729	2,300	-79%	1	1	0%	1,690	1,699	+1%	0	0	—
case-14 Our project directory contains an old `memory/board-meetings/` folder from an earlier software version. When writing new decision entries today, should we continue appending to that folder or transition to a different path?	fail→pass	8,093	2,816	-65%	1	1	0%	1,370	1,853	+35%	0	0	—
case-15 When a new board decision overrides an old policy from six months ago, should the old record file in Layer 2 be deleted to keep the folder clean?	fail→pass	10,080	3,529	-65%	1	1	0%	1,573	2,020	+28%	0	0	—
case-16 Where is the consolidated append-only index file of all approved decision records stored inside the Layer 2 directory structure?	fail→pass	11,977	2,059	-83%	1	1	0%	1,973	1,664	-16%	0	0	—
case-17 Who is responsible for checking conflicts and writing the approved records to Layer 2 after Phase 5 synthesis approval—do autonomous sub-agents write to Layer 2 directly?	fail→pass	11,092	2,792	-75%	1	1	0%	1,829	1,856	+1%	0	0	—
case-18 What python command line flag checks existing records for topic contradictions and DO_NOT_RESURFACE violations across our decision storage?	fail→fail	9,364	1,995	-79%	1	1	0%	1,503	1,722	+15%	0	0	—
case-19 If the founder modified the AI agent's recommended decision during synthesis, where in the standard decision entry format should this modification be documented?	pass→fail	6,389	2,478	-61%	1	1	0%	1,016	1,840	+81%	0	0	—
case-20 Can you prepare a 45-minute agenda for our upcoming product launch sync, including time allocations for engineering updates, marketing alignment, and Q&A?	pass→pass	9,499	6,417	-32%	1	1	0%	1,583	2,363	+49%	0	0	—
case-22 Please write a technical task description for updating our OAuth2 refresh token endpoint so our backend developer can copy it into Jira.	pass→pass	11,617	10,317	-11%	1	1	0%	2,171	3,133	+44%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +59 percentage points is the difference between those two pass rates over the 20 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 18.2%
Pass rate with the skill: 77.3%
Cases attempted: 22
Cases counted toward lift: 20
Cases that regressed: 2
Skill version measured: v1
Run completed: 8/11/2026, 09:54 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-04

Prompt

We want to store the complete unedited meeting debate from today's session—including all rejected agent arguments and counter-synthesis—into our raw memory archive. Where should this file be saved on disk, and what is the standard retention policy before moving it to historical archive folders?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run77

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+37% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 24 days ago

Other measured skills in the registry, with their headline benchmark lift.