New: 97,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Pact Engineer

by wandercom in wandercom/pact

Use Pact as a plan-first contract and test engine while the active Claude or Codex agent implements, reconciles infeasible contracts, validates the result, and runs Advocate plus Simulacrum review.

MITAuto-check passed

Install Pact Engineer

skills CLI
$ npx skills add wandercom/pact --skill pact-engineer -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install wandercom/pact pact-engineer --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/wandercom/pact.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/pact-engineer .claude/skills/pact-engineer && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
pact-engineer
GitHub stars
165
Token cost
~1.4k tokens
SKILL.md length
629 words
Files
4 (incl. references)
Skills in repo
2
Repo updated
First seen
Licence
MIT

At a glance

Use Pact as a plan-first contract and test engine while the active Claude or Codex agent implements, reconciles infeasible contracts, validates the result, and runs Advocate plus Simulacrum review.

  • Works in 6 steps: Source ~/.profile when feasible. → Start or resume a Kindex session, then… → Read repository instructions and the… → …
  • Substantial feature work
  • SKILL.md covers Workflow, Execution Modes, Production-Ready Minimalism and Review Tool Behavior
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Pact Engineer is an agent skill from wandercom/pact. Use Pact as a plan-first contract and test engine while the active Claude or Codex agent implements, reconciles infeasible contracts, validates the result, and runs Advocate plus Simulacrum review. Use for substantial feature work, architecture changes, risky fixes, or any task where the user asks to plan with Pact, run Pact plan-only, follow the engineer workflow, or obtain adversarial review before declaring done.

Its SKILL.md is about 1.4k tokens, which your agent loads only when the skill is triggered. The skill folder holds 5 other files, including reference files (for example `agents/openai.yaml`, `references/done-gate.md` and `references/lean-review.md`).

The repository describes itself as: Contracts before code. Tests as law. Agents that can't cheat. The licence is MIT.

When your agent uses it

  • Substantial feature work
  • Architecture changes
  • Any task where the user asks to plan with Pact
  • Run Pact plan-only

Example prompts

  • “/pact-engineer”

Workflow steps

6 steps, taken from the first numbered list in SKILL.md.

  1. Source ~/.profile when feasible.
  2. Start or resume a Kindex session, then search the task and affected domain.
  3. Read repository instructions and the relevant implementation, tests, and
  4. Normalize vague task language into a concrete problem statement,
  5. Run central architecture claims past Simulacrum before locking the frame.
  6. Run Pact in plan-only mode and advance it through interview/decomposition

What it can do on your machine

Read from SKILL.md and the folder at commit f27b38e. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are bash).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Pact Engineer loads about 1.4k tokens when it runs, and up to ~2.9k if it reads all its reference files. Until then it costs about 108 tokens; SKILL.md has 629 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~108
When it runs · the whole SKILL.md, loaded when a task matches
~1.4k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~2.9k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from wandercom/pact at commit f27b38e, republished under its MIT licence (© wandercom). 629 words, ~1,402 tokens.

Download SKILL.mdSave it as .claude/skills/pact-engineer/SKILL.md (or your agent's skills folder). This skill also uses 3 other files; get the full folder from GitHub.
name
pact-engineer
description
Use Pact as a plan-first contract and test engine while the active Claude or Codex agent implements, reconciles infeasible contracts, validates the result, and runs Advocate plus Simulacrum review. Use for substantial feature work, architecture changes, risky fixes, or any task where the user asks to plan with Pact, run Pact plan-only, follow the engineer workflow, or obtain adversarial review before declaring done.

Pact Engineer

Use Pact to make the implementation bet explicit before coding. Keep Pact's contracts and tests authoritative, but treat planning and implementation as a feedback loop rather than a one-way handoff.

Workflow

  1. Source ~/.profile when feasible.
  2. Start or resume a Kindex session, then search the task and affected domain.
  3. Read repository instructions and the relevant implementation, tests, and recent history before framing the change.
  4. Normalize vague task language into a concrete problem statement, boundaries, constraints, and acceptance criteria. Use Constrain artifacts when available; otherwise create equivalent prompt.md and constraints.yaml inputs.
  5. Run central architecture claims past Simulacrum before locking the frame.
  6. Run Pact in plan-only mode and advance it through interview/decomposition until it pauses after contracts and tests:
bash
pact init "$PACT_DIR" --budget 10
pact run "$PACT_DIR" --constrain-dir "$CONSTRAIN_DIR" --plan-only --once

Answer or approve interview questions as needed, then continue pact run --once until the plan-only pause. Do not bypass unresolved ambiguity.

  1. Read design.md, .pact/decomposition/tree.json, contracts, visible tests, and generated task lists. Run pact assess on the affected source tree.
  2. Render a focused brief before implementing each component:
bash
pact handoff "$PACT_DIR" "$COMPONENT_ID" --validate --max-tokens 2000
  1. Implement directly with the active agent. Follow the repository's existing patterns and verify incrementally.
  2. Reconcile contracts whenever implementation exposes infeasibility:
    • Stop instead of silently deviating.
    • Record the conflict and the failed assumption in Kindex.
    • Update task/constraints/contracts and regenerate affected tests or rerun Pact plan-only.
    • Resume only after the contract and implementation agree.
  3. Run repository tests, type checks, lint, pact validate, and pact audit where applicable. Then run the lean review checklist in references/lean-review.md: remove ceremony that does not improve correctness, recovery, observability, or evidence, but do not delete safety or production-readiness controls.
  4. Run post-implementation review:
bash
pact review "$TARGET_REPO" \
  --claim "This change is done because: <specific evidence>"

Treat Advocate critical/high findings and Simulacrum premise rejections as blocking. Fix the work, not merely the wording, then rerun review. 13. Check references/done-gate.md, capture durable decisions and results in Kindex, and end the Kindex session.

Execution Modes

  • Default to Pact plan-only plus active-agent implementation.
  • Use pact run <project> --implement only when the user explicitly wants Pact to own implementation and integration.
  • Use pact build <project> <component> for targeted Pact-managed implementation after plan-only decomposition.
  • After three genuine failed approaches or material scope expansion, stop, document the invalid assumption, and choose a simpler contract-satisfying approach or report the blocker.
Show full SKILL.md (260 more words)Show less

Production-Ready Minimalism

Use the lean lens as a separate review pass, not as a replacement for correctness, security, or production evidence.

  • Before adding code, stop at the first rung that holds: does this need to exist, does stdlib solve it, does the platform solve it, does an installed dependency solve it, can the minimum correct behavior be more direct?
  • New dependencies, abstractions, queues, brokers, services, wrappers, config layers, and extension points need a reason tied to a real requirement, measured constraint, or recovery/observability need.
  • Intentional shortcuts must state the ceiling, the trigger that makes the ceiling real, and the upgrade path.
  • For every layer, ask whether removing it makes a trust boundary unenforceable, a failure mode unobservable, or rollback/evidence gathering impossible, or turns a hard invariant into a manual discipline requirement. Yes to any means load-bearing; no to all four means ceremony.
  • Lean review may remove boilerplate, wrapper-only layers, speculative config, hand-rolled stdlib behavior, and unused flexibility.
  • Lean review may not remove trust-boundary validation, error handling that prevents data loss, security, accessibility, observability, rollback, migration safety, auditability, live validation, or evidence-backed gates.
  • The operational test: an operator must be able to answer what broke, why, and how to roll back without deploying new code.

Review Tool Behavior

pact review persists advocate.json, simulacrum.md, command output, and a machine-readable review.json under .pact/reviews/. Pact uses its packaged Simulacrum runtime and annotated corpus by default; PACT_SIMULACRUM_CMD is an explicit operator override. A successful command means requested tools completed and Advocate returned no critical/high findings; the active agent must still read and adjudicate Simulacrum's response.

© wandercom, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 3 other files (references) in skills/pact-engineer of wandercom/pact.

  • SKILL.md
  • agents/openai.yaml
  • references/done-gate.md
  • references/lean-review.md

Open the folder on GitHubat commit f27b38e

Compare with similar skills

Pact Engineer next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Pact Engineer compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Pact Engineer this skillwandercom/pact165—~1.4kAutomated safety check: PassMIT
Agentic Engineeringaffaan-m/ECC274k—~986Automated safety check: PassMIT
Prompt Engineeringdavila7/claude-code-templates32k7 repos~1.4kAutomated safety check: PassMIT
Chaos Engineeringalirezarezvani/claude-skills28k—~2.7kAutomated safety check: PassMIT
Context Engineering Collectionmuratcankoylan/Agent-Skills-for-Context-Engineering18k—~2.8kAutomated safety check: PassMIT
Prompt EngineerRightNow-AI/openfang18k—~829Automated safety check: PassApache-2.0

Similar skills

  • Agentic Engineering

    affaan-m/ECC

    Operate as an agentic engineer using eval-first execution, decomposition, and cost-aware model routing.

    274k GitHub stars~986 tokensUpdated 2 days ago
    AI & LLM EngineeringAuto-check passed

  • Prompt Engineering

    davila7/claude-code-templates

    Expert guide on prompt engineering patterns, best practices, and optimization techniques.

    32k GitHub starsUsed in 7 repos~1.4k tokens
    AI & LLM EngineeringAuto-check passed

  • Chaos Engineering

    alirezarezvani/claude-skills

    A skill your agent uses when planning, running, or learning from chaos engineering experiments.

    28k GitHub stars~2.7k tokensUpdated 1 mo ago
    DevOps & CloudAuto-check passed

  • Context Engineering Collection

    muratcankoylan/Agent-Skills-for-Context-Engineering

    A comprehensive collection of Agent Skills for context engineering, harness engineering, multi-agent architectures, and production agent systems.

    18k GitHub stars~2.8k tokensUpdated 6 days ago
    Agent WorkflowsAuto-check passed

  • Prompt Engineer

    RightNow-AI/openfang

    Prompt engineering expert for chain-of-thought, few-shot learning, evaluation, and LLM optimization

    18k GitHub stars~829 tokensUpdated 3 mo ago
    AI & LLM EngineeringAuto-check passed

  • Protocol Reverse Engineering

    wshobson/agents

    Master network protocol reverse engineering including packet analysis, protocol dissection, and custom protocol documentation.

    40k GitHub starsUsed in 7 repos~3.2k tokens
    SecurityAuto-check passed

More from wandercom/pact

  • Simulacrum

    wandercom/pact

    Stress-test an idea, plan, architecture frame, definition, criterion, draft, or done claim using Pact's packaged Jeremy Simulacrum.

    165 GitHub stars~313 tokensUpdated 3 days ago
    Auto-check passed

Questions about Pact Engineer

What does Pact Engineer do?

Use Pact as a plan-first contract and test engine while the active Claude or Codex agent implements, reconciles infeasible contracts, validates the result, and runs Advocate plus Simulacrum review. Pact Engineer is an agent skill from wandercom/pact. Use Pact as a plan-first contract and test engine while the active Claude or Codex agent implements, reconciles infeasible contracts, validates the result, and runs Advocate plus Simulacrum review.

When should I use Pact Engineer?

Pact Engineer fits situations like: substantial feature work; architecture changes; any task where the user asks to plan with Pact; run Pact plan-only.

How do I install Pact Engineer in Claude Code?

Run `npx skills add wandercom/pact --skill pact-engineer -a claude-code`. Or copy the skill folder (skills/pact-engineer in wandercom/pact) into .claude/skills/pact-engineer in your project. Claude Code loads it when a task matches its description.

How do I install Pact Engineer in Codex?

Run `npx skills add wandercom/pact --skill pact-engineer -a codex`. Or copy the skill folder (skills/pact-engineer in wandercom/pact) into .agents/skills/pact-engineer in your project. Codex loads it when a task matches its description.

Can I use Pact Engineer in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add wandercom/pact --skill pact-engineer -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/pact-engineer, .gemini/skills/pact-engineer, .github/skills/pact-engineer and .opencode/skills/pact-engineer in your project.

What does Pact Engineer need to run?

SKILL.md names no scripts, command-line tools or credentials: Pact Engineer is instructions for the agent only.

Does Pact Engineer access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Pact Engineer safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Pact Engineer use?

Pact Engineer is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Pact Engineer use?

About 1.4k tokens (SKILL.md is roughly 5.6k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 1.5k tokens, read only when the agent opens those files.

What are the alternatives to Pact Engineer?

Skills that share tags, products or a category with Pact Engineer: Agentic Engineering (affaan-m/ECC, 274k stars), Prompt Engineering (davila7/claude-code-templates, 32k stars), Chaos Engineering (alirezarezvani/claude-skills, 28k stars) and Context Engineering Collection (muratcankoylan/Agent-Skills-for-Context-Engineering, 18k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Pact Engineer?

wandercom (a GitHub organization) maintains it in wandercom/pact, which has 165 GitHub stars. The repository holds 2 skills in this directory. The repository was last updated on October 4, 2026.

Source: wandercom/pact on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.