Félagi

Put your Félagi board inside the session — issues, comments, time, the stopwatch, the knowledge base and agent runs, without leaving the terminal.

@elyracode/felagi talks to Félagi over its HTTP API with a personal token. Félagi is a project portal where people and agents share one board; this extension gives the agent on your machine the same view of the work that everyone else has.

Install

elyra install npm:@elyracode/felagi

Setup

Create a token in Félagi under Settings → API tokens. It hands back the block it is going to be pasted into — put it in your project's .env, or export the variables:

FELAGI_URL=https://felagi.example.com
FELAGI_TOKEN=fat_...

Tokens are not gated to admins: anybody who works there can make one, because the tool on your desk needs nobody's permission to read an issue. A read-only token works for everything except writes, which answer 403 with a message saying so.

Tools

Tool	Description
`felagi_me`	Who the token acts as, its workspace and abilities
`felagi_issues`	List and search issues: text, status, type, priority, project, assignee, labels, cycle, `updated_since`
`felagi_issue`	One issue in full, optionally with the comment thread and delivered artifacts
`felagi_create_issue`	Create an issue
`felagi_update_issue`	Change any subset of fields; assigning an agent queues a run
`felagi_comment`	Comment on an issue; mentioning an agent hands it the thread
`felagi_log_time`	Log time the way a person says it: `2h 30m`, `1d`, `90m`
`felagi_time_entries`	Read hours back, filtered by mine, date range, issue or source
`felagi_timer`	The stopwatch: status, start, stop, discard
`felagi_articles`	Search the team knowledge base and read full articles
`felagi_projects`	List projects
`felagi_agents`	List assignable agents with runtime status
`felagi_run`	Queue an agent run on an issue; list and read runs with summary and cost

Commands

Command	Description
`/felagi`	Dashboard — my issues, running timer, workspace
`/felagi-timer`	`/felagi-timer start ACM-231`, then `stop`, `status` or `discard`
`/felagi-log`	`/felagi-log ACM-231 2h30m traced the redirect loop`

Behaviour worth knowing

Issue references accept the human identifier (ACM-231) or a bare number — but they are not the same namespace. ACM-231 means issue number 231 in your workspace, with any prefix the workspace answers to and case ignored. A bare 231 is the database key, which is a different number. On an installation with more than one workspace, 231 and ACM-231 are different issues.
Prefer the prefixed form — from Félagi 0.44.0. It is what every response gives you, and it is stable across installations in a way the database key is not. Before 0.44.0 the API published ACM-231 as an issue's id and then refused it in a URL, so felagi_issue ACM-231 answered 404 against an older server; use the numeric key there. An unrecognised prefix is now a 404 rather than a quiet fall back to the number, so XYZ-231 never returns issue 231.
spent_on defaults to today, and the date recorded is when the work happened, not when it was entered.
Hours are always the token owner's own. The API has no actor field for time, deliberately: a credential that can log time against a colleague is one that can rewrite their timesheet.
Starting a timer while another runs stops and logs the first, and the tool reports what it banked. Switching tasks is normal; banking time against the wrong thing silently is not.
Time logged by the stopwatch is recorded as measured rather than recalled, and the time report leads with what proportion of a total was watched happening.

Why use it

The loop that happens every day — read what an issue says and record the work on it without switching to a browser, which is the friction that makes time tracking stop after a fortnight
The knowledge base is the endpoint worth having — skills are what an agent is given; felagi_articles is what it can look up, so an agent that reads the team's runbook first follows the team's conventions
Hand work to an agent from the terminal — assigning an agent queues a run, and felagi_run reads back its summary and cost
One board — the run, the comment and the hours land where the rest of the team already looks, rather than in a second system to reconcile

The API reference is at /docs/felagi/api, and Working from your editor is the guide this extension automates.