pipeline-state

pipeline-state is a skill for Claude Code from endorphin-ai/hasbrains-agent-kit. It costs 209 tokens per session (1,798 once invoked), scanned A, original, MIT.

A file-based shared record for a multi-agent software workflow. It stores run details, phase results, and links to evidence files in a separate JSON file for each session.

In plain words
What is it for?
Use it to pass small result fields and evidence locations between an orchestrator and worker agents without copying large outputs into every prompt.
Why use it?
It prevents important handoff information from being lost when conversations become long or are compressed.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the context-engineering plugin — 3 skills shipped together

Good fit Use it to pass small result fields and evidence locations between an orchestrator and worker agents without copying large outputs into every prompt.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/endorphin-ai/hasbrains-agent-kit/pipeline-state
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add endorphin-ai/hasbrains-agent-kit --skill pipeline-state
Clone the repo
git clone --depth 1 https://github.com/endorphin-ai/hasbrains-agent-kit

Made for: Claude Code.

Or install context-engineering, the plugin that ships this one along with the rest of its 3 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for pipeline-state

README.md
[![agentmods](https://agentmods.dev/badge/skills/endorphin-ai/hasbrains-agent-kit/pipeline-state/github.svg)](https://agentmods.dev/skills/endorphin-ai/hasbrains-agent-kit/pipeline-state)
Your own site
<a href="https://agentmods.dev/skills/endorphin-ai/hasbrains-agent-kit/pipeline-state"><img src="https://agentmods.dev/badge/skills/endorphin-ai/hasbrains-agent-kit/pipeline-state/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for pipeline-state

Your own site · 80×15
<a href="https://agentmods.dev/skills/endorphin-ai/hasbrains-agent-kit/pipeline-state"><img src="https://agentmods.dev/badge/skills/endorphin-ai/hasbrains-agent-kit/pipeline-state.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 209 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,798 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00209 $0.01798
Opus 5 $0.00105 $0.00899
Sonnet 5 $0.00042 $0.00360
Haiku 4.5 $0.00021 $0.00180

Measured 10d ago against content hash 8beea16076c5, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

pipeline-state scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.

The scan reads SKILL.md. This mod also ships 1 executable file (scripts/pipeline-state.sh), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

context-engineering/skills/pipeline-state/SKILL.md · 127 lines

How it starts

The opening of the file, as written. The whole thing — 127 lines — stays where its author put it; the contents beside it link to each section on GitHub.

pipeline-state — the run-wide handoff bus

Durable, file-backed state for the ENTIRE pipeline. One JSON file persists the run's variables + per-phase outputs + .ai_log/ evidence-path references across phases, so a downstream agent (or the orchestrator) recovers exactly what it needs by LOOKUP instead of receiving a large inlined dump — and nothing is lost to context compression on long runs.

The standing pattern: evidence -> .ai_log/; references + key fields -> pipeline-state; handoffs = state lookups, not inlined dumps.

  • the orchestrator INITIALIZES the state at session start (alongside creating the session folder), passes its path (state_file) to every dispatched agent, and after each phase READS that phase's output fields + .ai_log/ evidence paths FROM the state to validate the gate and to build the NEXT phase's prompt — passing state keys + paths, never re-dumping large content.
  • every agent, on START, READS the state to pull the inputs this phase needs (key fields + .ai_log/ evidence paths) — it does NOT expect large inlined content; on COMPLETION, WRITES its outputs — key result fields + PATHS to evidence offloaded in .ai_log/ (never the blobs).
  • It COEXISTS with the docs/-native reports (the human-readable record) and the .ai_log/ path-handoff. The state REFERENCES the docs/ + .ai_log/ artifacts; it does not replace them.

State file (PER-SESSION, .ai_log/, git-ignored)

.ai_log/session-<session-id>-<name>.json

ONE file per run, living under .ai_log/ — e.g. .ai_log/session-2026-06-30-member-blog-gating.json.

Naming convention. <session-id> is the run's disambiguating id — commonly the session date <YYYY-MM-DD> (it separates repeat/concurrent runs); <name> is the epic slug. Together <session-id>-<name> is the session LABEL <YYYY-MM-DD>-<epic-slug> (the same slug as session_dir). The script DERIVES the path from that label (session-<label>.json).

The orchestrator derives + creates it at init and passes its exact PATH as state_file to every dispatched agent. Agents read/write the path they were given — they do NOT reconstruct or hardcode a constant path, because it varies per session. Use the one convention consistently across the skill, the orchestrating command, and any workflow definition.

Read the full file on GitHub · 127 lines

Files

What ships with it

3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 10d ago First seen · 127 lines · 0 tokens per session scan A 8beea16076c5

Subscribe to this mod's changes

pipeline-state is a skill published in the GitHub repository endorphin-ai/hasbrains-agent-kit (4 stars, last pushed 1mo ago), licensed MIT. It adds 209 tokens to every session and 1,798 once invoked, about $0.0010 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

create-task

Use this skill to crystallise and scope an ongoing conversation into a well-defined task, capturing user intent at a high level. Typical workflow: create-task → clarify-task → plan-task → impl-task → review-code → review-task → ship-task.

patforna/auto-task · 52 tokens

quality-stack

Scan a project to detect configured quality and testing tools across JVM (Gradle/Maven), Android (AGP/Compose/KMP), Node.js/TypeScript, and Python ecosystems. Cross-reference against research-backed recommendations and assist with setup. Auto-detects project type(s) including monorepos with mixed ecosystems. Use when…

joaquimscosta/arkhe-claude-plugins · 148 tokens

lyra

Transform vague inputs into precision-optimized AI prompts for Claude, ChatGPT, Gemini, or other LLMs. Use when user mentions "optimize prompt", "improve prompt", "lyra", "prompt engineering", or needs help crafting effective AI prompts.

joaquimscosta/arkhe-claude-plugins · 56 tokens

prototype

Rapid UI prototyping — generates 3 visually distinct HTML/CSS components from a text prompt and serves them in a live browser gallery. Use when user runs /prototype, asks to "prototype a UI", "mock up a component", "generate HTML mockups", "create UI variations", "design exploration", "quick UI concept", or wants to…

joaquimscosta/arkhe-claude-plugins · 118 tokens

clarify-task

Use this skill to resolve ambiguities, clarify vague criteria and make implicit assumptions explicit. Invoked at the end of /at:create-task and the start of /at:plan-task. Typical workflow: create-task → clarify-task → plan-task → impl-task → review-code → review-task → ship-task.

patforna/auto-task · 65 tokens

sops-add-key

Add a new machine's age public key to .sops.yaml and re-encrypt all files. Use for multi-machine setups. Use when user mentions "add key", "add machine", "sops add key", "new machine", "authorize machine", "share key", "add public key", "multi machine sops".

joaquimscosta/arkhe-claude-plugins · 72 tokens