pi-code@1.10.1
VERIFIED· install onlyPublisher claims: "Claude Code experience for the pi coding agent: reads your .claude config (rules, commands, skills, hooks, output styles, MCP servers, agents) and adds todo, checkpoints, memory, web, subagents, and goals" Observed in sandbox: no network, touches nothing outside the project, writes no memory.
Coverage
Nothing executable in the bundle ran; the trace covers the install step only.- Tested
- every file read, file written and host contacted by what ran
- Not tested
- use of the skill by an agent
What changed since 1.7.0
vet-skills diff pi-code 1.7.0 1.10.1This version reaches for something 1.7.0 did not: 3 new hosts (agentskills.io, claude.com, duckduckgo.com) · 2 new credential variables (MCP_CLIENT_SECRET, MCP_OAUTH_CALLBACK_PORT). Compared from the two recorded runs, not from source.
- New hosts
- agentskills.io, claude.com, duckduckgo.com
- New credential variables
- MCP_CLIENT_SECRET, MCP_OAUTH_CALLBACK_PORT
The guard can hold an update like this until someone approves it (runtime.on_change: hold). A change that only shows with a real credential, or on a path the run did not reach, is not in a trace and so not in this comparison.
Is [email protected] safe to install?
KillerSkills ran it in the KillerSkills honeypot sandbox (1 scenario; nothing executable in the bundle ran; the trace covers the install step only) on 2026-11-21 and scored it 81/100 (Verified). In what ran, nothing we could detect went wrong: no undeclared network egress, no credential reads, no persistent instruction writes. What it does with a real credential was not observed; the static findings and the publisher carry the rest of the score. Summary: no network, touches nothing outside the project, writes no memory.
Behavior trace
Honeypot run · 0 s · 1 scenarios · 0 deviationsStatic findings
// that would kill the command is not passed on.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.* repo's `CLAUDE.md` cannot read `~/.claude` (`.credentials.json`, global settings, everyFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.// `if: Edit(~/.ssh/*)` guard miss the very call it names. The value is normalisedFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.* ~/.ssh, ../.. traversal, symlinks) is ignored, so a hostile CLAUDE.md, or aFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.run.kill = undefinedFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.* next turn, shown to nobody else. A timeout kill discards the output, like aFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.run.kill = () => {Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.run.kill = undefinedFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.* `onChild` hands the caller a kill for the spawned tree, so a background hook that isFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.// A process the shell forked while the kill was landing survives it and holds theFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.* the process-group kill used by the shell runner.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell."biome": "npx --no-install @biomejs/biome check",Fix: Pin versions and sources; vendor anything fetched at install time.* of auto memory, so the same settings chain and env kill switch gate it. */Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.* directory, read/write/curation instructions, and its MEMORY.md capped like theFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct./** Above 2^31-1 ms Node clamps a timer to 1ms, which would kill the hook instantly. */Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.if (ctx.hasUI !== true) for (const kill of backgroundKills) kill()Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell./** Claude's maxTurns: kill the child once it has produced this many turns. */Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.…p its structure and voice and make targeted edits.`) } else { parts.push('Write the result to AGENTS.md at the project root. Create the file yourself. pi prefers AGENTS.md over CLAUDE…Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.const kill = (): void => killProcessTree(child, 'SIGKILL')Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.// timeout can kill the descendants too. CLAUDE_PROJECT_DIR is Claude's documented way for a hook toFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.// slot with no kill and no eviction for the life of the process.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.// Its own process group off Windows, so the kill reaches what the shell spawned.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.kill = registeredFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell."knip": "npx --no-install knip"Fix: Pin versions and sources; vendor anything fetched at install time.killInflight = killFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell."biome:fix": "npx --no-install @biomejs/biome check --write .",Fix: Pin versions and sources; vendor anything fetched at install time.* the chmod then makes it exact where the umask took bits away. */Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.Cost to the agent
Tokens estimated at four characters each. Skills load their instruction text when triggered; the figure is what that costs.
Known vulnerabilities
No advisories in OSV for pi-code@1.10.1 as of 39 min ago. This is the package itself, not its dependencies.