Guide · Agent skill
How to use the accint-commitments agent skill: triage open promises
Updated · written with AI from the project's own documentation and checked against our catalogue data
accint-commitments is an agent skill that teaches an AI agent to review the open promises (commitments) tracked by the acc system, decide which ones can be closed, and record whether each was actually fulfilled. It is written for developers who already run the acc (accreted intelligence) MCP tools. By the end of this guide you will know what the skill does, how to install it, how to ask for it in natural language, and what to check before letting it close anything.
Commitment Triage for accsickn33/agentic-awesome-skills · MIT Download the skill Quick start
- 1Install acc. Make sure the acc command-line tool and the acc MCP server with the acc_act tool are set up and working.
- 2Download the ZIP. Download the accint-commitments skill ZIP from this page.
- 3Unzip into a skills folder. Unzip into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
- 4Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP.
- 5List open commitments. Ask the agent to triage your open promises, which runs acc commitments to list them.
- 6Verify before closing. Check the evidence for each candidate, then let the agent record verdicts with acc_act using the outcome runtime.
- 7Leave waiting items open. Keep genuinely pending commitments open rather than closing them early.
What does the commitments skill make the agent do?
The skill covers one job: triaging the open promises that acc tracks for an agent. It lists those open items, works out which ones can be closed, and records a verdict for each through an acc_act tool call.
The skill is explicit that its verdicts should be honest. It tells the agent to label its own grades accurately, and to reserve stronger credit for verdicts confirmed by an owner or by real-world evidence, such as a passing test or a real reply. Commitments that are still genuinely waiting can stay open, and the skill treats that as a valid state.
- Lists open commitments with the acc CLI command acc commitments (read-only).
- Records a verdict for each closeable item with acc_act(runtime="outcome").
- Distinguishes self-graded verdicts (weak signal) from owner-validated or reality-confirmed ones.
- Leaves commitments that are still waiting open.
acc commitmentsInstall the accint-commitments skill
Download the skill ZIP from this page. Then choose one of two locations. For a personal install available across your projects, unzip it into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder.
If you use the Claude apps instead, upload the ZIP under Settings → Capabilities → Skills.
- Personal: unzip into ~/.claude/skills/
- Project only: unzip into .claude/skills/ inside the project
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Check the requirements before you start
The skill cannot work on its own. It depends on two pieces of the acc system: the acc command-line tool, and an acc MCP server that exposes the acc_act tool with the outcome runtime. If those are missing, the commands and tools the skill names will not exist.
The skill runs no scripts and uses no network access, according to its own description. Everything it does goes through acc.
- acc command-line tool installed
- acc MCP server exposing the acc_act tool with runtime "outcome"
- No bundled scripts; the package contains SKILL.md only
How to trigger the skill with natural requests
You do not need a special command. Ask the agent in plain language to triage your commitments. The skill is meant to be picked up when the request matches its purpose, so phrasing close to its description works best.
- "Triage acc's open promises and close the ones that are done."
- "List my open acc commitments and tell me which ones I can close."
- "Review the open commitments and only close the ones you can verify."
- "Which of my open commitments are still waiting? Leave those open."
Example of a verdict call
For each commitment the agent decides is closeable, the skill uses the acc_act tool with the outcome runtime. The fields are a reference id, a good flag set to true or false, and a note. Provenance matters: the default self_graded grade is a weak signal and counts at 0.25×. Pass owner only when the owner has validated the result, and external or runtime only when reality confirmed it, such as a real reply or a passing test. The agent should never tag its own grade as reality.
- ref: the commitment id from the list
- good: true if fulfilled, false if not
- note: a short explanation of the evidence
acc_act(runtime="outcome", input={"ref": "<id>", "good": true, "note": "..."})Tips for safe, useful triage
Check reality before closing anything. Closing a commitment that was not really done, by marking your own work as complete, can distort the record. Review the evidence first, and keep waiting items open rather than forcing a verdict.
Read the verdicts the agent proposes before they are recorded, especially when the stakes are high. The skill's own limitations note says to verify commands and behaviour in your environment and to get user approval for consequential actions.
- Ask the agent to show the evidence for each verdict before it records it.
- Prefer owner or external confirmation over self-graded verdicts.
- Leave commitments open when the work is still in progress.
What people use it for
- Cleaning up a backlog of open commitments an agent made in earlier sessions.
- Recording an owner-validated outcome after a human has reviewed the work.
- Closing a commitment when a passing test or a real reply confirms it was fulfilled.
- Auditing whether an agent's stated promises were actually kept.
Frequently asked questions
Is the accint-commitments skill free?
The skill is published under the MIT licence, as stated on this page. It still requires the acc system, which this page does not price or describe further.
How do I install the accint-commitments skill?
Download the ZIP and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, upload the ZIP under Settings → Capabilities → Skills.
What do I need to run it?
You need the acc command-line tool and an acc MCP server that exposes the acc_act tool with the outcome runtime. Without acc, the commands and tools the skill names will not exist.
Does the skill run scripts or use the network?
According to its description, it runs no scripts and uses no network access. It consists of a SKILL.md file only.
Why does the agent say a verdict is weak?
Verdicts the agent grades itself are the default self_graded type and are treated as a weak prior, credited at 0.25×. Stronger credit needs owner validation or real-world evidence.
Who should not use this skill?
Anyone without the acc system installed should not, because the commands and tools will not exist. It is also a poor fit if you want to close commitments without checking reality first.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill is well suited to developers already running acc who want an agent to clean up its open commitments with careful, evidence-based verdicts. It is not useful without the acc CLI and MCP server, and if you just want general task tracking, pick a tool that does not depend on acc.
Download the skill