codex-history-ingest reads the session logs Codex CLI keeps under ~/.codex and compiles the durable knowledge in them into an Obsidian wiki. It starts from the session index, parses the rollout files, and skips token counts and tool plumbing.
Findings are clustered by topic across sessions, written with summaries, confidence fields and provenance markers, and recorded in a manifest so later runs only handle new or changed sessions.
What You Get
- Wiki pages grouped by project and topic, distilled from Codex sessions
- A manifest entry per rollout file so later runs process only the delta
- Redaction of API keys, tokens and private identifiers before anything is written
- Provenance markers and a summary line on each page
Common Use Cases
- Turn months of Codex CLI work into a searchable vault
- Sync only the sessions run since the last ingest
- Recover recurring debugging techniques from past rollouts
- Add Codex history to a vault that already holds Claude history
What Makes It Unique
It is written for Codex’s own layout: a session index with thread names, dated rollout files made of event envelopes, and an optional history file. It names which event types carry knowledge and which are noise, and it treats injected system prompts as material that must never be copied into the wiki.
Under The Hood
The skill is instructions for the agent. It tells the agent to read the vault config and manifest, list and read files under the Codex history folder, and call the obsidian-wiki command line to sync the index, log and hot cache. It writes markdown pages and manifest entries into your vault. If a QMD collection is configured it runs the qmd command to refresh the search index. It makes no web requests of its own.