Skip to content
Skills Library
AI & agents · Testing

langgraph-proof

/langgraph-proof

Use when building, reviewing, or debugging a LangGraph flow that must prove a resumed run preserved its recorded semantic transitions and side-effect discipline.

Licence
Apache-2.0
Last repository change
27.10.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 27.10.2026, without references to retired models or APIs.

Licence

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

What it does and when to use it

Use when building, reviewing, or debugging a LangGraph flow that must prove a resumed run preserved its recorded semantic transitions and side-effect discipline.

Original author description

Use when building, reviewing, or debugging a LangGraph flow that must prove a resumed run preserved its recorded semantic transitions and side-effect discipline.

How it works

  1. Start with the boundary
  2. What to prove
  3. Interpret the result precisely
  4. MCP use
  5. Unified native audit operation
Original SKILL.md headings · 87 lines, 486 words
SKILL.md preview
---
name: langgraph-proof
description: Use when building, reviewing, or debugging a LangGraph flow that must prove a resumed run preserved its recorded semantic transitions and side-effect discipline.
---

# Code Factory LangGraph Proof

Use Code Factory as a local proof layer around a team-owned LangGraph test
harness. It is not a graph runtime, checkpoint store, repair agent, or release
authority.

## Start with the boundary

Before relying on the MCP server, check that the current workspace has the
Code Factory CLI and inspect its stated local boundary:

```bash
factory --version
factory mcp status --root .
```

If the CLI is missing, explain that the user must install
`factoryline-code-factory>=0.43.0` in the environment that will host the local
MCP process. Do not install Python packages, change a client configuration, or
start a server without the user's approval.

## What to prove

For a LangGraph flow with a durable checkpoint or interruption path, ask for:

1. A team-owned reference harness run.
2. A separately captured resumed or interrupted run.
3. Two sealed, workspace-relative transition lineages. The recorder retains
   SHA-256 facts about state and side-effect identifiers, not raw state,
   prompts, or secrets.

Compare existing lineages without writing a file:

```bash
factory langgraph replay-verify \
  --root . \
  --reference .factory/langgraph/reference.json \
  --resumed .factory/langgraph/resumed.json \
  --json
```

Only add `--out .factory/langgraph/assurance.json` after the user explicitly
approves writing a local assurance receipt.

## Interpret the result precisely

- `LANGGRAPH_RESUME_PARITY_VERIFIED` means the supplied sealed lineages match
  semantically and the deterministic analysis found no supported anomaly.
- `LANGGRAPH_REPLAY_DIVERGENCE` means the supplied evidence differs or exposes
  duplicate completed effects, stale reads/writes, or unsafe parallel writes.
  Present the incident capsule and recovery cone for human review.
- `LANGGRAPH_INPUT_REJECTED` means the comparison was rejected before a
  receipt was written; fix the supplied inputs rather than guessing.

Never claim that a parity receipt proves production resilience, complete
instrumentation, external idempotency, savings, quality, or correctness beyond
the supplied lineages.

## MCP use

`factory.langgraph_assurance` is read-only. It may compare two existing
workspace-relative receipts; it cannot invoke a graph, mutate checkpoints,
replay effects, write a receipt, approve a repair, merge, deploy, publish,
send a message, or access credentials.

If a divergence needs code changes, prepare a narrow repair proposal and tests
for a human to approve. Do not authorize or execute repairs from the incident
capsule.

## Unified native audit operation

Use `factory scan --root PATH --deep --json` for the local audit overview.
For authorized native execution, add `--worker-config PATH` with signed,
SHA-pinned worker configuration. Select workers for the repository languages
and risk: CodeQL Python, JavaScript/TypeScript, or GitHub Actions; Semgrep's
Python and JavaScript/TypeScript rules; dependency, secret, configuration,
runtime and fuzz workers as applicable. MCP status tools read evidence;
they do not launch these workers.

Bind findings and repair instructions to candidate, profile, tool and image
hashes. Retain native SARIF flows and reproductions. Rerun affected checks after
a repair, then obtain separate specialty AI review. Reuse evidence only when
its complete candidate and execution bindings match; an inventory or stale
receipt is never a pass. Missing runtime, mutation or language coverage stays
incomplete. Avoid launching every heavyweight worker for each edit.

Versions

The database has no other skills named langgraph-proof.

Database snapshot: 13.11.2026 · package version d2eb4d9542f4 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100llm-application-dev-langchain-agent
    Build production-grade AI agents with LangChain and LangGraph. Use when designing, implementing, or debugging a LangChain or LangGraph agent.
  • 100langgraph-workflows
    Build agent workflows with LangGraph 1.0 state machines and graph patterns. Use this skill when creating agent graphs, implementing state machines, building multi-step agent workflows, or using LangGraph. Activate when: LangGraph, agent graph, state graph, agent workflow, graph…
  • 100langgraph
    Use when building or debugging LangGraph workflows — designing state graphs, adding conditional routing, wiring checkpointers, streaming tokens, implementing human-in-the-loop interrupts, or coordinating multi-agent subgraphs.
  • 100langgraph
    Use when building or debugging LangGraph workflows — designing state graphs, adding conditional routing, wiring checkpointers, streaming tokens, implementing human-in-the-loop interrupts, or coordinating multi-agent subgraphs.
  • 100langgraph
    LangGraph expertise for building stateful, multi-step agent workflows. Use when the user asks about LangGraph, StateGraph, nodes, edges, conditional routing, checkpointers, persistence, memory, human-in-the-loop, subgraphs, streaming, or building agents with langgraph /…
  • 100langgraph
    Expert in LangGraph, the production-grade framework for building stateful, multi-actor AI applications. Use when building agent graphs, managing state, or implementing persistence with checkpoints.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.