npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

agoragentic-premortem-golden-loop

v0.1.7

Published

Free local OSS premortem, self-heal, and no-spend Golden Loop readiness CLI for plans, launches, and installable AI agent repositories.

Vulnerabilities

Powered byPowered by Snyk
  • 0High
  • 0Medium
  • 0Low

Links

Git

Maintainers

rhein1rhein1

Keywords

agoragenticagent-osai-agentspremortemgolden-loopself-healx402receiptsoss

Readme

Agoragentic Premortem Golden Loop Agent

npm version license node engine

Agoragentic Premortem Golden Loop Agent

OSS premortem agent for plans, launches, products, hires, strategies, and installable AI agent repositories. It can generate a full six-month failure-frame premortem report, run a repo release premortem, check the local no-spend Golden Loop, propose safe self-healing fixes, and write machine-readable receipts that an owner can inspect before publishing, deploying, or enabling paid execution.

This standalone repo is the canonical npm source for agoragentic-premortem-golden-loop (also vendored as a copy inside agoragentic-integrations/premortem-golden-loop).

Agoragentic family

| Repo / package | What it is | |---|---| | agoragentic-integrations | 93 public integration surfaces across frameworks, protocols, SDKs, commerce rails, and governance tools | | agoragentic-ecf-core | Self-hosted context-governance runtime (npm agoragentic-ecf-core) | | Micro ECF | Open local context wedge (npm agoragentic-micro-ecf) | | fable5-codex | Evidence-first Codex audits, reviews, fact checks, and repo sweeps | | agoragentic-premortem-golden-loop (this repo) | Pre-launch release-readiness CLI (npm agoragentic-premortem-golden-loop) | | agoragentic-summarizer-agent | Python example: route summarize via execute() | | agoragentic-openai-agents-example | OpenAI Agents SDK marketplace example |

Home: agoragentic.com · all packages: npm view <name>

Agent workflow contracts: governed agent runs and Fable review output.

This package is local-first by default:

  • free to use
  • no Agoragentic API key required
  • no wallet required
  • no repo contents, prompts, business plans, reports, or receipts sent anywhere
  • no network calls unless explicitly requested
  • no paid execution
  • no production mutation
  • self-heal never overwrites existing files, deletes files, rotates secrets, deploys, or publishes

Install

Published on npm. Run it directly with npx, no install or clone required:

npx agoragentic-premortem-golden-loop doctor --repo .
npx agoragentic-premortem-golden-loop audit --repo .

More commands:

npx agoragentic-premortem-golden-loop run --repo .
npx agoragentic-premortem-golden-loop serve --repo . --host 127.0.0.1 --port 8787

Develop from source

From a local clone of this repository:

npm test
node bin/agoragentic-premortem-golden-loop.mjs run --repo ../my-agent

Integration recipes and ready-to-copy templates live in docs/INTEGRATIONS.md, docs/EXTERNAL_AGENT.md, and templates/. The repo includes IDE agent snippets, GitHub Actions, Docker/home-server examples, a dependency-free MCP stdio server, and an opt-in HTTP external-agent server.

Release steps live in docs/RELEASE.md. Sanitized sample outputs live in examples/.

Commands

# Explain what the agent will read/write before running the audit.
node bin/agoragentic-premortem-golden-loop.mjs doctor --repo .

# One-command local audit: repo premortem, Golden Loop, HTML guide, heal plan, IDE handoff.
node bin/agoragentic-premortem-golden-loop.mjs audit --repo .

# Full audit with business context so the premortem report is complete.
node bin/agoragentic-premortem-golden-loop.mjs audit --repo . \
  --plan "Launch an OSS AI agent that runs premortems and Golden Loop readiness checks" \
  --audience "AI agent builders preparing public releases" \
  --success "builders install it, run it, and make one concrete launch change"

# Apply only missing additive scaffolds after reviewing the audit.
node bin/agoragentic-premortem-golden-loop.mjs audit --repo . --apply-safe-fixes

# Rerun the audit to close the loop. The report tracks what was applied,
# what is now verified resolved, and what remains open.
node bin/agoragentic-premortem-golden-loop.mjs audit --repo .

# Full Klein-style premortem session for a plan, launch, or decision.
node bin/agoragentic-premortem-golden-loop.mjs session \
  --plan "Launch an OSS AI agent that runs premortems and Golden Loop readiness checks" \
  --audience "AI agent builders preparing public releases" \
  --success "builders install it, run it, and make one concrete launch change"

# Full local premortem plus no-spend Golden Loop receipt.
node bin/agoragentic-premortem-golden-loop.mjs run --repo .

# Self-heal plan only. No files are changed.
node bin/agoragentic-premortem-golden-loop.mjs heal --repo .

# Apply safe additive fixes: missing docs, agent.json, .env.example, or CI scaffold.
node bin/agoragentic-premortem-golden-loop.mjs heal --repo . --apply-safe-fixes

# Static repo release premortem only.
node bin/agoragentic-premortem-golden-loop.mjs premortem --repo .

# Golden Loop readiness only.
node bin/agoragentic-premortem-golden-loop.mjs golden-loop --repo .

# CI mode: fail when blockers or Golden Loop failures remain.
node bin/agoragentic-premortem-golden-loop.mjs run --repo . --ci

# Explicit offline mode. This is also the default.
node bin/agoragentic-premortem-golden-loop.mjs run --repo . --skip-network

# Optional public no-spend canaries. Sends no repo contents.
node bin/agoragentic-premortem-golden-loop.mjs run --repo . --allow-network-canaries

# Optional runtime probe for a locally running agent.
node bin/agoragentic-premortem-golden-loop.mjs run --repo . --target-url http://localhost:3000

# Optional external HTTP agent for private tools that cannot speak MCP.
node bin/agoragentic-premortem-golden-loop.mjs serve --repo . --host 127.0.0.1 --port 8787

Artifacts are written to:

.agoragentic/premortem-golden-loop/
  premortem-report-[timestamp].html
  premortem-transcript-[timestamp].md
  premortem-session-[timestamp].json
  doctor.json
  doctor.md
  audit.json
  audit-guide.html
  audit-summary.md
  closure-loop.json
  closure-loop.md
  ide-fix-prompt.md
  agent-handoff.md
  premortem.json
  premortem.md
  golden-loop.json
  golden-loop.md
  local-receipt.json
  summary.md
  healing-plan.json
  healing-plan.md
  healing-recheck.json

Workflow For Users

Premortem Golden Loop workflow

  1. Paste the GitHub repo into a local IDE/LLM and ask it to run npx agoragentic-premortem-golden-loop doctor --repo ..
  2. Review the doctor output: it explains what will be read, what local artifacts will be written, and what the agent will never do.
  3. Run audit --repo . to produce the local repo premortem, Golden Loop receipt, HTML guide, self-heal plan, and IDE/agent handoff prompts.
  4. If the business context is missing, rerun audit with --plan, --audience, and --success so the full premortem report can be generated.
  5. Give ide-fix-prompt.md or agent-handoff.md to a local IDE agent to implement safe fixes from the findings.
  6. Run audit --repo . --apply-safe-fixes only after reviewing the plan; this creates missing additive scaffolds and still does not delete or overwrite code.
  7. Rerun audit --repo .; the closure loop compares the prior local audit with the current repo and writes closure-loop.md / closure-loop.json.
  8. Rerun audit --repo . --ci; optionally add --run-tests if the repo's declared tests are safe in no-spend mode.
  9. Use Agent OS, Micro ECF, x402, hosted deployment, marketplace publication, or paid execute() only as separate owner-approved steps.

One-Command Audit Flow

audit is the intended default for local IDEs and other coding agents. It combines:

  • doctor: the safety and consent gate
  • repo premortem: local release and operating risk scan
  • no-spend Golden Loop: install/config/discovery/proof/approval/test readiness
  • premortem session: full HTML report when plan, audience, and success context are available
  • launch gate: source files read, assumptions refused, risky actions blocked, and the exact IDE prompt handed off
  • closure loop: tracks applied safe fixes, prior recommendations now verified resolved, and still-open actions across local reruns
  • self-heal plan: safe additive implementation plan
  • IDE handoff: prompts another local agent can use to fix Golden Loop readiness without destructive changes

By default, audit writes only local artifacts. With --apply-safe-fixes, it may create missing scaffold files, but it still will not delete files, overwrite existing files, edit application source code, install dependencies, deploy, publish, call paid execute(), sign wallets, or transfer funds.

Integrations

See docs/INTEGRATIONS.md and docs/EXTERNAL_AGENT.md for exact setup steps.

External HTTP Agent

serve is an opt-in HTTP surface for local/private agents that cannot use stdio MCP. It binds to 127.0.0.1 by default:

npx agoragentic-premortem-golden-loop serve --repo . --host 127.0.0.1 --port 8787

Available endpoints include GET /health, GET /.well-known/agent.json, GET /tools, and POST /audit. Non-loopback binding requires AGORAGENTIC_EXTERNAL_AGENT_TOKEN or --external-agent-token; remote safe fixes, network probes, and test execution require separate owner-approval flags. See docs/EXTERNAL_AGENT.md.

Examples And Release

Sanitized sample artifacts:

Release checklist:

Automatically validated:

  • syntax checks
  • Node test suite
  • local no-spend run --repo . --ci
  • local no-spend audit --repo . --ci --skip-network
  • npm package dry run during local release validation

Manual owner checks/actions:

  • Docker runtime build/run, when Docker is available
  • MCP client configuration inside the target client
  • external HTTP agent smoke test on the target host or private network
  • npm publish
  • Git tag and GitHub release
  • GitHub social preview upload using assets/social-card.png

Premortem Session Workflow

The session command implements the prompt in PROMPT.md:

  • checks whether it has the minimum context: what it is, who it is for, and what success looks like
  • frames the plan as already failed six months from now
  • generates the raw failure reasons
  • runs one independent investigator pass per failure reason
  • synthesizes the most likely failure, most dangerous failure, hidden assumption, revised plan, and pre-launch checklist
  • writes a self-contained dark HTML report and a full Markdown transcript

If context is missing, it writes premortem-context-needed.json and asks the next single question instead of producing a generic report.

What The Premortem Checks

The premortem looks for release blockers and operating risks:

  • README, OSS license, and reproducible install contract
  • declared test contract
  • agent discovery metadata such as agent.json, agent-card.json, SKILL.md, OpenAPI, MCP, or a manifest
  • committed secret-like values without printing the secret value
  • .env.example or equivalent configuration instructions
  • explicit no-spend, budget, owner approval, x402, USDC, or paid-execution boundaries
  • receipt, trace, invocation, reconciliation, or audit-proof contract
  • basic runtime operations notes such as health, readiness, rollback, or runbook
  • Agent OS / Micro ECF / execute(task,input,constraints) alignment when the repo is meant to launch through Agoragentic

What The No-Spend Golden Loop Tests

The local Golden Loop is a readiness loop, not a settlement proof:

  1. install contract exists
  2. configuration and secret boundary is clear
  3. agent discovery contract exists
  4. premortem blockers are resolved
  5. receipt/proof contract exists
  6. owner approval and spend boundary is explicit
  7. public no-spend Agoragentic canaries respond, only when --allow-network-canaries is used
  8. optional target runtime responds, when --target-url is provided
  9. optional declared tests pass, when --run-tests is used

The public canaries are off by default. If enabled, they use only unauthenticated no-spend surfaces and do not send repository contents:

  • GET /api/discovery/check
  • GET /api/x402/info
  • GET /api/x402/test/echo
  • GET /api/catalog?spend_possible=false&auth=none

Self-Healing Boundaries

The heal command is deliberately conservative. In plan mode it writes only healing-plan.json and healing-plan.md under .agoragentic/premortem-golden-loop/.

With --apply-safe-fixes, it may create only missing additive scaffolds:

  • docs/AGORAGENTIC_GOALS.md
  • docs/AGORAGENTIC_WORKFLOWS.md
  • docs/AGORAGENTIC_SAFETY_BOUNDARIES.md
  • agent.json
  • .env.example
  • .github/workflows/agoragentic-premortem-golden-loop.yml

It does not overwrite existing files, edit application source code, delete files, remove secrets, rotate credentials, install dependencies, call paid execute(), transfer USDC, publish listings, deploy, or open a production runtime. Secret findings and license choices remain manual owner actions.

Paid Proof Boundary

This package intentionally does not sign wallet payments or run paid execute() calls. Real Golden Loop proof on Agoragentic includes wallet ownership, funding, quote-backed execution, receipt, and reconciliation. That path must remain explicitly owner-approved and budget-gated.

For integrations, keep external paid work routed through:

execute(task, input, constraints)

Do not hardcode provider IDs unless the agent intentionally needs a specific provider.

Repository Images

Deterministic brand assets live in assets/. The SVG sources are generated by scripts/generate-brand-assets.mjs, then exported to PNG for GitHub and README surfaces.

Run npm run assets:generate to regenerate SVG sources. Use docs/IMAGE_PROMPTS.md only if creating new raster variants. Keep images consistent with the repo promise: local-first, free, no data sent anywhere by default, and explicit owner approval before any paid or networked path.