agentleFS
krmisystemskrmisystems/fantasy-football-manager/skills/espn-automation/SKILL.md

espn-automation

Skill91 starsChanged 2 months agoMITView on GitHub

Connect the ESPN companion for browser drafts or HTTP season operation. Run continuous analysis and submit exact transactions within saved user limits.

The file

Show its rules and text

We split this skill into the rules it gives an agent

29 rules. About 0 are common, 8 are distinctive, and 21 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 21

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use the fantasy-football-espn companion supplied by this plugin for live ESPN drafts.

    line 9
  2. Very specific:

    Use season-manager for HTTP lineups, acquisitions, waiver claims, drops, and IR through the same companion.

    line 10
  3. Distinctive:

    Use the manager MCP for saved configuration and strategy.

    line 11
  4. Distinctive:

    Check the current release evidence before claiming live account acceptance.

    line 14
  5. Very specific:

    Read get_capabilities, get_manager_config, and espn_get_status.

    line 20
  6. Very specific:

    Call espn_connect with that context and phase="draft".

    line 22
  7. Very specific:

    Call espn_sync to obtain a complete current board.

    line 24
  8. Very specific:

    Read espn_get_status before starting submissions.

    line 25
  9. Distinctive:

    Do not request passwords or cookies through tool arguments.

    line 32
  10. Very specific:

    Use the season-manager workflow for the exact season action and its required evidence.

    line 37
  11. Very specific:

    Never send credential values through tool arguments or prompts.

    line 40
  12. Very specific:

    Call espn_sync before analysis.

    line 45
  13. Distinctive:

    Use explicit browser season mode only when that workflow is authorized.

    line 48
  14. Distinctive:

    Use separate state directories for separate team contexts.

    line 50
  15. Very specific:

    Use the user's authorized mode and limits.

    line 58
  16. Very specific:

    Use update_manager_config with the current revision for an authorized policy change.

    line 59
  17. Very specific:

    Call espn_start_automation for continued observations and simulation batches.

    line 62
  18. Very specific:

    Only the draft_pick action mode automatic permits the loop to select and submit picks itself.

    line 63
  19. Distinctive:

    Do not request per-pick confirmation when the user has already authorized automatic mode.

    line 64
  20. Very specific:

    Use espn_start_standalone_worker when the user requests operation after Codex closes.

    line 68
  21. Distinctive:

    Read status before starting another worker.

    line 73
  22. Very specific:

    Call espn_stop_automation to pause new actions across processes sharing the data directory.

    line 75
  23. Very specific:

    Use espn_disconnect to release this process's connection after its work ends.

    line 77
  24. Very specific:

    Use espn_prepare_draft_pick(player_id) to refresh the board and prepare a specific pick.

    line 82
  25. Very specific:

    Use espn_submit_draft_pick(proposal_id, confirmation=true) after required review confirmation.

    line 84
  26. Very specific:

    If a response reports awaiting_verification, call espn_reconcile_draft_pick.

    line 88
  27. Very specific:

    Never retry an uncertain submission as a new click or proposal.

    line 89
  28. Very specific:

    Read espn_get_status for worker health, pending claims, and current recommendations.

    line 102
  29. Distinctive:

    Do not claim championship odds or unverified live write acceptance.

    line 104

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

ESPN automation

Use the fantasy-football-espn companion supplied by this plugin for live ESPN drafts. Use season-manager for HTTP lineups, acquisitions, waiver claims, drops, and IR through the same companion. Use the manager MCP for saved configuration and strategy. The published v0.4.0 preview adds HTTP season operation and exposes 16 ESPN tools. Version 0.3.3 retains 13 ESPN tools, browser drafts, and one-swap lineup proposals. Check the current release evidence before claiming live account acceptance. Distinguish installed-package checks, authenticated reads, live submissions, and fictional HTTP or browser tests. Trade execution remains unavailable.

Connect and verify a draft

  1. Read get_capabilities, get_manager_config, and espn_get_status.
  2. Identify the requested league ID, team ID, and season.
  3. Call espn_connect with that context and phase="draft".
  4. Have the user complete ESPN sign-in in the browser when required.
  5. Call espn_sync to obtain a complete current board.
  6. Read espn_get_status before starting submissions.

Draft phase uses the existing browser adapter and a dedicated local Chrome profile. The source browser extra supplies Playwright. Installed Chrome is also required. The old draft Chrome extension is not required. It does not copy cookies from an existing profile. The optional cdp_url must identify a loopback debugging endpoint. Do not request passwords or cookies through tool arguments. ESPN Autopick must be verified disabled before direct submission.

Connect a season without a browser

Use the season-manager workflow for the exact season action and its required evidence. HTTP mode requires a protected session file with only SWID and espn_s2. Configure the file path through FFM_ESPN_CREDENTIAL_FILE or the server --credential-file option. Never send credential values through tool arguments or prompts. The optional Linux v10 importer reads an existing authorized session without starting Chromium. It does not supply password sign-in or renew an expired session.

Connect with phase="season", transport="http", and the requested week. Call espn_sync before analysis. Verify account ownership, targeted roster locks, current period, and pending claims. HTTP operation requires neither a browser process nor Playwright. It rejects cdp_url. Do not use browser control as an implicit fallback. Use explicit browser season mode only when that workflow is authorized.

Use separate state directories for separate team contexts. Both MCP servers for one context must use the same directory. HTTP controllers sharing an account must use one credential directory to share team leases. The same league and team cannot have two active controllers through that directory. Different teams have independent leases.

Continuous operation

Use the user's authorized mode and limits. Preserve unrelated configuration fields. Use update_manager_config with the current revision for an authorized policy change. A strategy change does not grant submission authority.

Call espn_start_automation for continued observations and simulation batches. Only the draft_pick action mode automatic permits the loop to select and submit picks itself. Do not request per-pick confirmation when the user has already authorized automatic mode. Review mode requires exact proposal confirmation as described below.

The ordinary loop runs inside the MCP process. Use espn_start_standalone_worker when the user requests operation after Codex closes. The standalone worker uses saved policy. Draft mode stops when the selected roster is complete. HTTP season mode follows the explicit week unless authorized auto_rollover=true enables verified current-period changes. Unresolved submissions and pending waivers retain their original week. Backward or unverified periods stop rollover. Legacy browser season mode requires an explicit reconnect for another week. Read status before starting another worker.

Call espn_stop_automation to pause new actions across processes sharing the data directory. Clear automation.paused through an authorized configuration update when resuming. Use espn_disconnect to release this process's connection after its work ends. Disconnect is not a shared pause for another worker.

Exact picks and verification

Use espn_prepare_draft_pick(player_id) to refresh the board and prepare a specific pick. Show its player, pick number, league, mode, and required confirmation. Use espn_submit_draft_pick(proposal_id, confirmation=true) after required review confirmation. Automatic mode can submit without per-pick confirmation within the user's authorization.

A proposal can grant only one click claim. If a response reports awaiting_verification, call espn_reconcile_draft_pick. Never retry an uncertain submission as a new click or proposal. Missing, empty, or failed browser results do not prove success. Report confirmed only after the platform identifies the expected player at the expected pick. Report not_selected when that pick identifies another player.

For HTTP season actions, use espn_prepare_season_action, espn_submit_season_action, and espn_reconcile_season_action. Authorization records one durable claim before its POST. Repeated submission cannot send another request for that proposal. pending_waiver does not establish ownership. A dependent lineup change must wait for confirmed acquisition. not_submitted identifies a proved preflight failure before the transaction request. Correct its cause before preparing another authorized proposal. An uncertain result requires reconciliation without a new submission or replacement proposal. Named coverage repairs require saved player authorization. Unknown projections must remain unknown.

Read espn_get_status for worker health, pending claims, and current recommendations. Report actual trials and source age. Treat model scores and availability as estimates. Do not claim championship odds or unverified live write acceptance. Report the actual transport, action modes, unresolved results, and verified scope of automatic operation.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.