Follow the chain of ancestor sessions each session was compacted from, up
to --max-depth (default 10), summarizing each older session more tightly
via a per-depth decay factor. Guards against cycles and already-visited
sessions. Bounds the full chain to ~24k tokens worst case.
Default to the most recently active session in the current project,
excluding the caller's own session (found via CLAUDE_CODE_SESSION_ID).
Lets you run /clear then cc-compact to reload the just-cleared session,
emulating Claude Code's built-in /compact.
Vendored the caveman-compress skill from JuliusBrussee/caveman and
renamed it to compact-file. Compresses natural language memory files
(CLAUDE.md, todos) to save input tokens; preserves code, URLs, paths,
and structure, with a verified out-of-tree backup.
Claude-Session: https://claude.ai/code/session_01B8oPygSSYuomdr8pKxLwko
Gathers a GitHub PR's title, description, reviews, per-line comments, and
conversation comments via gh into one XML report (gather_pr.py), tagging each
author as is_me/is_bot/is_human and giving every comment an id. The skill reads
that output, evaluates each review point, fixes obvious bugs/security issues,
asks the user on nuanced calls, replies to bots directly, and confirms with the
user before replying to humans.
Claude-Session: https://claude.ai/code/session_01AhP6LpRrKxH7yJPb6pfLFE
Companion to cc-compact: resolves a session by id or name, then searches
across user prompts, agent replies, commands, file ops, tool calls, and
tool output via regexp, glob, or ranked keyword matching. Bounded output
(centred snippets, --limit/--truncate tuned to ~2k tokens). cc-compact now
points to it for closing concrete gaps the compact report leaves open.
- cc-compact: skill to reload a past Claude Code session's context
(resolves by id or ai-title, extracts a bounded XML summary via
compact_session.py) so work can be resumed without ingesting the log
- migrate the existing new-work skill into the repo
- deploy ~/.claude/skills via ensure_dir_symlink in setup.sh
- claude-token-count: count tokens via Anthropic's count_tokens endpoint,
pulling the API key from 1Password at runtime
