pi-history-ingest

Extracts durable knowledge from Pi coding agent session JSONL files into an Obsidian wiki.

Updated Nov 9, 2023
One-click install
npx skills add https://github.com/oresttokovenko/dot-files --skill pi-history-ingest-oresttokovenko
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pi-history-ingest
Source: https://github.com/oresttokovenko/dot-files/tree/main/chezmoi/dot_agents/skills/pi-history-ingest
Command: npx skills add https://github.com/oresttokovenko/dot-files --skill pi-history-ingest-oresttokovenko

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Pi coding agent sessions accumulate as raw JSONL transcripts that are hard to revisit, so the knowledge, decisions, and workflows captured in past sessions remain buried and unusable. ## Core Features & Use Cases - Session Parsing: Reads Pi session JSONL files, reconstructs the active conversation branch from the tree structure, and extracts messages, compaction summaries, and bash executions. - Incremental Ingest: Tracks processed files in a .manifest.json so only new or modified sessions are re-processed, with a full re-ingest mode available. - Wiki Distillation: Clusters extracted knowledge by topic and writes distilled pages into an Obsidian vault with provenance markers, source reference comments, and confidence metadata. - Use Case: After weeks of Pi coding sessions across several projects, run the ingest to surface recurring debugging patterns and project decisions as searchable wiki pages with auditable source citations. ## Quick Start Ask the assistant to process your Pi history and add your Pi sessions to the wiki.

Frequently Asked Questions about pi-history-ingest

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I import my Pi agent session history into Obsidian?

Point the skill at your Pi sessions directory (default ~/.pi/agent/sessions) and it parses each JSONL file, extracts durable knowledge, and writes distilled pages into your Obsidian vault. It updates a manifest so subsequent runs only process new or changed sessions.

What data does Pi session JSONL parsing extract?

It extracts user and assistant messages, compaction summaries, branch summaries, and bash execution commands from the active conversation branch. It skips thinking blocks, model change events, labels, and extension state entries as noise.

Does the ingest handle sensitive data like API keys in session logs?

Yes, the skill applies a privacy filter that redacts API keys, tokens, passwords, and credentials before writing anything to the wiki. It summarizes bash outputs containing paths or environment variables rather than quoting them verbatim.

Can I re-ingest all Pi sessions instead of only new ones?

Yes, full mode processes every session file regardless of the manifest, which is useful after a wiki rebuild. The default append mode only processes files that are new or modified since their recorded ingestion time.

Why are source reference comments added to wiki pages?

Each extracted claim carries a source comment pointing to the specific session file and entry ID it came from, creating an audit trail. A verification gate checks every claim against this evidence ledger before any page is written.