feat: OOB secure input improvements + credential_store toolset#128
Merged
feat: OOB secure input improvements + credential_store toolset#128
Conversation
…E}} resolution in execute_command Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ls; move SEC_RE to module scope
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
password import was being mistakenly used as a variable — replaced all references with inputValue. Removed unused @inquirer/password import. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
kumaakh
force-pushed
the
feat/oob-improvements
branch
from
8765a04 to
6dda598
Compare
getOrCreateKey() was writing raw binary to FLEET_DIR/key, but the test (and original design) expects a hex-encoded 32-byte key at FLEET_DIR/salt. Changed to read/write hex at SALT_PATH so the persistent key file is human-readable and matches the expected path.
…rompt The OOB credential collection terminal showed "Enter API key" and "API key:" labels, which are misleading when collecting arbitrary secrets via credential_store_set. Renamed both to "Secure Value".
Extends the same token resolution logic from execute-command to
register_member so that credentials stored via credential_store_set
can be referenced as {{secure.NAME}} in the password parameter.
H1: apply resolveSecureTokens() to restart_command in execute-command.ts H2: apply redactOutput() to long-running task launch output M1: credentialDelete now removes from both session and persistent tiers unconditionally M2: OOB confirmation prompt uses plain visible readline input instead of masked secureInput M3: pass input.prompt through collectOobApiKey() → auth.ts --prompt arg → secureInput M4: add 16 tests covering credential round-trip, token resolution, output redaction, network egress policies L1: remove unused KEY_PATH, SALT_LENGTH, getOrCreateSalt from crypto.ts L2: change "API key received" to "Secure value received" in auth.ts L3: add "Best-effort heuristic — not a security boundary" comment on NETWORK_TOOL_RE
…et skill, PM skill
Add early guard in execute_prompt that returns an error if the prompt
contains {{secure.NAME}} tokens — secrets must never reach LLM context.
Fix docs in SKILL.md, tpl-doer.md, onboarding.md, and README.md to
clarify that {{secure.NAME}} resolution only happens in execute_command
and specific MCP tool params, never in execute_prompt prompts.
This was referenced Apr 20, 2026
Closed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
New tools
{{secure.NAME}} token resolution
Tokens resolved server-side — LLM never sees plaintext. Output redacted ([REDACTED:NAME]) including monitor_task output. Network egress policy (allow/confirm/deny) with OOB TTY confirmation.
Review
Reviewed by fleet-rev — APPROVED. All 9 findings addressed (2 HIGH, 4 MED, 3 LOW).
🤖 Generated with Claude Code