csift

Solid

Read, search and analyze Claude Code session transcripts (the .jsonl under ~/.claude/projects), and message another running lane. Use this INSTEAD of grep/ripgrep/cat/jq/python: the format has documented traps that return plausible wrong answers, no error (a user-role filter overcounts human turns 3x; a sixth of human turns hide inside tool_result payloads, AskUserQuestion answers included). Reach for it when you would hand-scan a session jsonl or shadow session facts in a state file - and BEFORE asserting something was never done or re-deriving a harness mechanism from binaries: past sessions usually derived it already. Search any regex across ALL sessions; read records by line, turn or uuid; what a session is doing now; tools, tokens, models, files changed; extract pasted images; recover a deleted file or plan; restore the turns a compaction clipped. Message another session or subagent the official tool cannot reach. Sub-second, hook-safe, pure regex, not semantic search; writes only its own sidecar.

Data & Documents 31 stars 0 forks Updated today MIT

Install

View on GitHub

Quality Score: 82/100

Stars 20%
50
Recency 20%
100
Frontmatter 20%
70
Documentation 15%
100
Issue Health 10%
80
License 10%
100
Description 5%
100

Skill Content

# csift — ripgrep for Claude Code session transcripts Surface: **v0.12.1** (must == `csift --version`). MECHANICAL GUARD: at first use after any compaction or context restore, run `csift --version` — if it differs from this Surface line, the copy you are reading is a stale in-context echo, and the installed SKILL.md is the one that matches the binary: Read it before anything else. Same diagnosis when an invocation you were CONFIDENT about errors (an older surface from prefill/summary/habit); never fall back to hand-parsing the jsonl. Rust CLI over CC session `.jsonl` under `~/.claude/projects/<encoded-cwd>/`. Built for an LLM consumer: token-lean text, uniform JSON, pure regex (RE2-class, linear-time; no backrefs/lookaround — they fail to compile by design). Smart-case: a pattern is case-insensitive unless it carries an uppercase; `-i` forces insensitive. `csift <cmd> --help` is the authoritative flag manual. Flag order is genuinely free — before/after the subcommand, before/after positionals, all equivalent. ## Why not hand-roll this format Every row below is measured, not hypothetical, and none of them threw an exception: a hand-written pass returns a plausible number that is wrong in a direction you cannot see. | what a hand-written pass does | what it actually returns | the csift move | |---|---|---| | filters `type:"user"` for human turns | **3.03x overcount** (3,607 vs 1,191 corpus-wide; 8.6x in the most multi-agent lane): peer-session inbox messages and harness no...

Details

Author
wdhwg001
Repository
wdhwg001/csift
Created
3 weeks ago
Last Updated
today
Language
Rust
License
MIT

Similar Skills

Semantically similar based on skill content — not just same category

AI & Automation Listed

past-sessions

Search and read past agent sessions on this machine — your own and other agents' — with agentcarto. Use when the user refers to earlier work ("what did we decide", "carry on from last time", "when did we touch this file"), or when the context you need is in a session rather than in the code.

2 Updated 1 weeks ago
agentcarto
AI & Automation Listed

session-intelligence

Analyze the user's own local Claude Code, Codex, and Cursor coding-session history to find evidence-backed workflow patterns and safely turn approved findings into rules, skills, commands, agents, docs, notes, or statistics. Use for requests such as analyze my sessions, review how I use AI, find repeated instructions, reduce token or context waste, improve my prompts, discover documentation gaps, compare Claude/Codex/Cursor usage, mine successful workflows, measure rework or verification, check whether a rule helped, or convert a recurring pattern into a durable artifact.

0 Updated 1 months ago
mluev
AI & Automation Listed

harvest

Mines Claude Code session transcripts for the corrections that earn artifacts, the intake stage of the kitchen's loop. Extracts what the user actually said across recent sessions (via a bundled script that reduces multi-MB transcripts to user turns), dispatches readers to classify corrections, redirects, and repeated failures, dedups against the standing harness, and hands surviving candidates to the forge with evidence counts. Use on a cadence (end of a heavy week, after a project push), when a correction feels familiar, or before claiming a harness has no gaps; skip when recent sessions were light or single-topic.

3 Updated 1 months ago
claudialnathan