recall 0.3.0

Secrets stay out of other agents' context.

Claude Code memories sometimes capture a credential from a debugging session. Served over MCP, that text would reach another agent's model provider. From 0.3.0, memory_get, memory_recall, memory_list and memory_due replace recognisable secrets with [redacted] before returning them, and the reply says how many were hidden. Files on disk are unchanged.

• Covered: passwords in connection URLs (scheme, user and host kept), bearer tokens, private key blocks, Anthropic, OpenAI, GitHub, Slack, AWS, Google and Stripe keys, and JWTs. Ordinary text (model ids, ticket numbers, password-less URLs) is left alone.
• Turn it off with --allow-secrets or RECALL_ALLOW_SECRETS=1.
• README: a tested Cline setup line next to Codex, Cursor and Gemini CLI (#2).

Suggested in openai/codex#58220. Changes: #1, #2.

Install or update:

npx -y @fainir/recall@0.3.0 --claude-code /path/to/your/repo --read-only
gemini extensions update recall

recall 0.2.1

Packaging release, same code as 0.2.0.

• Gemini CLI: gemini extensions install https://github.com/fainir/recall now works (0.2.0 was tagged before the extension manifest existed). It serves the Claude Code memory of the folder you start Gemini in.
• npm (npx -y @fainir/recall) and the MCP Registry entry stay at 0.2.0: the package contents are identical.

v0.2.0

recall 0.2.0 is now on npm as @fainir/recall and listed in the official MCP Registry as io.github.fainir/recall.

codex mcp add recall -- npx -y @fainir/recall --claude-code /path/to/your/repo --read-only

Reads and writes Claude Code's per-project auto memory (plain Markdown with wiki-links) so Codex, Cursor and other MCP clients share the same memories. Zero dependencies.