Use when the user asks to finish or wrap up the session — checks that all they asked for is delivered, then commits, pushes and archives it. Not for a code review. Self-contained — execute the inlined steps directly, do NOT invoke other skills.
79%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Critical
Do not install without reviewing
Fix and improve this skill with Tessl
tessl review fix ./plugins/flowai/skills/finish-session/SKILL.mdDetected a prompt injection in the skill instructions. The skill contains hidden or deceptive instructions that fall outside its stated purpose and attempt to override the agent’s safety guidelines or intended behavior.
The skill is a disclosed session-closing workflow with strict gates, refused force pushes, and a required user answer before any push, but its Reflect Phase autonomously edits the project's instruction and skill files and commits them without asking, which alters the user's work beyond the finish task.
The skill prompts the agent to compromise the security or integrity of the user’s machine by modifying system-level services or configurations, such as obtaining elevated privileges, altering startup scripts, or changing system-wide settings.
The skill performs operational changes (pushing to the remote, which can trigger CI, and editing tracked agent instruction files), but it fails the irreversible-loss gate because the destructive `git reset --hard HEAD~1` is guarded and reversible, `git rm -r` is recoverable from history, and archiving requires confirmation for uncommitted files.
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
The required workflow reads CI run logs via the AGENTS.md `Logs command` on the CI-red handoff path (line 243), and those logs can echo content from the pushed commit that an outsider may have influenced, such as test names or output; the workflow fetches them only after a selected failure, so this is low rather than medium. The remaining inputs (user messages, git status/diff/log, repo docs, task files) are first-party or the user's own content.
korchasa/flowai-plugins
b8c8990
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.