feat(skills): improve agent-tty skill score 85% to 99%#121
Open
rohan-tessl wants to merge 2 commits into
Open
Conversation
Hey @ThomasK33 👋 really like how the bootstrap skill delegates to the CLI as the source of truth while still being discoverable by agents. The "Playwright for terminals" framing in the README nails the positioning, and the Codex/Claude demo videos showing real agents using the tool are a nice touch. I ran your skill through `tessl skill review` at work and found some targeted improvements. Here's the before/after: | Skill | Before | After | Change | |-------|--------|-------|--------| | agent-tty | 85% | 99% | +14% | the description was already scoring perfectly at 100%. The content score had the most headroom so I focused there. <details> <summary>Changes made</summary> **Content improvements (biggest impact):** - **Quick start workflow**: added a complete create, run, wait, capture, destroy example so agents can start using `agent-tty` immediately without needing to run `agent-tty skills get` first - **TUI automation example**: added a second example showing the `--no-wait` plus `wait --screen-stable-ms` plus `send-keys` pattern for interactive TUIs - **Command surface reference**: added a compact listing of all command groups (lifecycle, input, observe, capture, environment) so agents know what's available at a glance - **Preserved bootstrap design**: kept the delegation to `agent-tty skills get agent-tty` for the full canonical skill with extended patterns and anti-patterns, just moved it to a clear "Full Skill and Extensions" section **What stayed the same:** - Description (100%, untouched) - Frontmatter structure and advertise flag - All three `agent-tty skills get/list` references </details> I also stress-tested your `agent-tty` skill against a few real-world task evals and it held up really well on driving an interactive nvim session with keystrokes and capturing proof artifacts. Kudos for that. quick honest disclosure. I work at https://github.com/tesslio where we build tooling around skills like these. Not a pitch, just saw room for improvement and wanted to contribute. If you want to self-improve your skills, or define your own scenarios to pressure test, just ask your agent (Claude Code, Codex, etc.) to evaluate and optimize your skill with Tessl. Ping me @rohan-tessl, if you hit any snags. Thanks in advance 🙏
Member
|
Hey @rohan-tessl, appreciate the PR. The quick start guide and removing the fact that the CLI is the source of truth regarding skills defeats the purpose of a bootstrap skill, as it softens the requirement to load the skill via CLI. Also, the real-world task evals you linked are not related to the agent-tty skill, but rather to Matt Pocock's to-issue skill, which is imported and vendored in this project via the skills cli. |
Hey @ThomasK33 👋 really like how the bootstrap skill delegates to the CLI as the source of truth while still being discoverable by agents. The "Playwright for terminals" framing in the README nails the positioning, and the Codex/Claude demo videos showing real agents using the tool are a nice touch. I ran your skill through `tessl skill review` at work and found a small, targeted improvement. Here's the before/after: | Skill | Before | After | Change | |-------|--------|-------|--------| | agent-tty | 85% | 88% | +3% | the description was already scoring perfectly at 100%. The content score had some headroom, specifically on progressive disclosure, so I focused there while keeping the bootstrap minimal. <details> <summary>Changes made</summary> **Content improvements (small, additive):** - **Core loop hint**: added a one-liner naming the create, run, wait, capture, destroy loop so agents understand the workflow shape before loading the full skill - **Command scope summary**: added a brief listing of command categories (lifecycle, input, observation, capture) and the `--json` / `--home` flags so agents know what's available at a glance - **Preserved bootstrap design**: kept the "source of truth" statement, kept the skill minimal, kept all three `agent-tty skills get/list` references unchanged - Used fenced code blocks for the CLI commands (slightly easier to copy) **What stayed the same:** - Description (100%, untouched) - Frontmatter structure and advertise flag - The intentional bootstrap philosophy </details> quick honest disclosure. I work at https://github.com/tesslio where we build tooling around skills like these. Not a pitch, just saw room for improvement and wanted to contribute. If you want to self-improve your skills, or define your own scenarios to pressure test, just ask your agent (Claude Code, Codex, etc.) to evaluate and optimize your skill with Tessl. Ping me @rohan-tessl, if you hit any snags. Thanks in advance 🙏
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
2 participants
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.
Hey @ThomasK33 👋
really like how the bootstrap skill delegates to the CLI as the source of truth while still being discoverable by agents. The "Playwright for terminals" framing in the README nails the positioning, and the Codex/Claude demo videos showing real agents using the tool are a nice touch.
I ran your skill through
tessl skill reviewat work and found some targeted improvements. Here's the before/after:the description was already scoring perfectly at 100%. The content score had the most headroom so I focused there.
Changes made
Content improvements (biggest impact):
agent-ttyimmediately without needing to runagent-tty skills getfirst--no-waitpluswait --screen-stable-msplussend-keyspattern for interactive TUIsagent-tty skills get agent-ttyfor the full canonical skill with extended patterns and anti-patterns, just moved it to a clear "Full Skill and Extensions" sectionWhat stayed the same:
agent-tty skills get/listreferencesI also stress-tested your
agent-ttyskill against a few real-world task evals and it held up really well on driving an interactive nvim session with keystrokes and capturing proof artifacts. Kudos for that.quick honest disclosure. I work at https://github.com/tesslio where we build tooling around skills like these. Not a pitch, just saw room for improvement and wanted to contribute.
If you want to self-improve your skills, or define your own scenarios to pressure test, just ask your agent (Claude Code, Codex, etc.) to evaluate and optimize your skill with Tessl. Ping me @rohan-tessl, if you hit any snags.
Thanks in advance 🙏