Files
worklane/assets/TOOL.ts
T
Ansho ReiandClaude Opus 5 11145b228e (c092ea7) Ship the opencode tool definitions with the binary, behind install-tool
`.opencode/tools/worklane.ts` was the second agent asset and the only one
that existed nowhere but a checkout. `cargo install` places binaries and
nothing else, so anyone who installed worklane got the skills and never the
tools, and the copy that did exist was reachable only from this repo. It is
now `assets/TOOL.ts`, embedded with `include_str!` like every other asset,
and `install-tool` writes it where opencode looks.

Its own command rather than a flag on `install-skill`, because the two
artifacts are not alike. A skill is inert markdown and one path serves both
agents. The tool is executable TypeScript and serves one: Claude Code has no
directory it scans for a tool definition, `claude mcp add` is the only way a
tool enters its list, and its plugin system adopts agents, hooks, monitors,
workflows, output-styles and themes and nothing named tools. Reaching it
would mean an MCP server, which is a different artifact and not this file.
So somebody on Claude Code takes the documents and never the wrapper, which
is the choice this command exists to give them.

The path was measured rather than guessed, as the ticket asked. 1.15.10
loads `~/.config/opencode/tool/` and `~/.config/opencode/tools/` both: each
was served in turn from a directory outside any project and
`/experimental/tool/ids` came back with all eleven worklane tools. Singular
is the target, the spelling opencode's own documentation puts first the way
it writes `agent(s)` and `skill(s)`. `tools_root()` follows
`XDG_CONFIG_HOME` where `skills_root()` does not, because opencode's config
root is an XDG path and `~/.claude` is not; `opencode debug paths` reports
`$XDG_CONFIG_HOME/opencode` with it set. An empty value counts as unset, the
convention `colors_wanted` already follows.

`opencode_tool_dir` takes both variables as arguments for the reason
`colors_wanted` does: a threaded test binary cannot mutate the process
environment to reach one branch at a time.

The file name is load bearing and now pinned by a test. opencode builds each
tool id as the file name then the export, so `worklane.ts` plus eleven
`export const` lines is what makes them `worklane_add` through
`worklane_uncheck`. Renaming the file renames every tool and every
permission rule naming one.

The closing line is not the skills' "on their next session". opencode reads
its config once at startup and never reloads it, so a running one keeps the
tools it began with and the honest instruction is to restart it.

Nothing else has to be installed, which is worth saying because the file is
executable and reads as though it wants a toolchain. Criterion 6 asked the
command to name the plugin package and Bun as runtime requirements and both
are wrong: a copy in a project with no `node_modules` above it registered
all eleven tools on a machine with no `bun` on `PATH`. opencode is a Bun
binary and the import resolves inside it. The message says the opposite, and
the reasoning is on the ticket.

`install_each` is the outcome loop both commands now share, with the
compare-and-refuse invocation passed in: that line is the only part of an
outcome that has to name the command it came from. What each command says
after the loop stays with the command.

`every_tool_the_wrapper_exports_is_a_real_subcommand` guards the direction
that breaks at call time, a tool wrapping a command that has been renamed or
dropped. Deliberately not the reverse: the wrapper has no tool for `relate`,
`campaign`, `assign` or `tags`, so an agent holding the tools cannot reach
campaigns or relations at all, and a test asserting that gap would fail on a
decision nobody has taken rather than on a mistake. Its own ticket.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-23 14:14:14 +02:00

480 lines
17 KiB
TypeScript

import { tool, ToolContext } from "@opencode-ai/plugin";
import path from "path";
function resolveWorklaneDir(context: ToolContext): string {
return path.join(context.worktree || context.directory || ".");
}
/**
* Run worklane and return its stdout, throwing with its stderr on failure.
*
* Both streams have to be piped. Bun defaults stderr to inherit, so it went to
* this process and never reached the model, and Bun.$ throws a ShellError whose
* message is only "Failed with exit code 1". Either way the CLI explained itself
* and nothing relayed the explanation.
*
* stdout is the fallback, not the source: worklane writes errors to stderr.
*/
async function run(cmd: string[], context: ToolContext): Promise<string> {
const proc = Bun.spawn(["worklane", ...cmd], {
cwd: resolveWorklaneDir(context),
stdout: "pipe",
stderr: "pipe",
});
const [stdout, stderr, exitCode] = await Promise.all([
new Response(proc.stdout).text(),
new Response(proc.stderr).text(),
proc.exited,
]);
if (exitCode !== 0) {
// anyhow prefixes its own "Error: ", and every caller here adds a prefix of
// its own, so drop it rather than relaying "Error adding comment: Error: ...".
const reason = stderr.trim().replace(/^Error:\s*/, "");
throw new Error(reason || stdout.trim() || `worklane exited with code ${exitCode}`);
}
return stdout.trim();
}
/**
* Description lines as worklane wants them, one --description value per line,
* blank ones included.
*
* cmd_add joins the values with a single newline, so an empty value is how a
* paragraph break is spelled. Dropping the blanks here, which is what the old
* filter(Boolean) did, collapsed every multi paragraph description into a single
* paragraph, against this tool's own instruction to separate paragraphs with
* newlines.
*
* Only the ends are trimmed, and only trailing whitespace within a line. Leading
* whitespace is content: it is what makes a nested markdown list or an indented
* block survive the trip.
*/
function descriptionLines(text: string): string[] {
const lines = text.split("\n").map((line: string) => line.trimEnd());
let start = 0;
let end = lines.length;
while (start < end && lines[start] === "") start++;
while (end > start && lines[end - 1] === "") end--;
return lines.slice(start, end);
}
/** One value per non-empty line, for the arguments that hold a list of items. */
function itemLines(text: string): string[] {
return text
.split("\n")
.map((line: string) => line.trim())
.filter(Boolean);
}
export const init = tool({
description:
"Initialize worklane in the current project. Creates the .worklane/ directory with backlog, committed, reviewing, and done subdirectories.",
args: {},
async execute(_args: Record<string, any>, context: ToolContext) {
try {
return await run(["init"], context);
} catch (err: any) {
return `Error initializing worklane: ${err.message}`;
}
},
});
export const add = tool({
description:
"Add a new ticket to the worklane backlog. Requires a title. Description, acceptance criteria, and plan each support multiple lines (one item per line). Tags are comma-separated.",
// Title is the only required field. Every other flag is left off entirely when
// the field is omitted, so an unmarked schema only forced the model to invent a
// value for fields it had nothing to say about.
args: {
title: tool.schema
.string()
.describe("Ticket title (required)"),
description: tool.schema
.string()
.describe("Description text. Use newlines between paragraphs.")
.optional(),
acceptance: tool.schema
.string()
.describe("Acceptance criteria, one per line. Put each criterium on its own line.")
.optional(),
plan: tool.schema
.string()
.describe("Implementation plan steps, one per line. Put each step on its own line.")
.optional(),
tags: tool.schema
.string()
.describe("Comma-separated tags, e.g. 'bug,frontend,urgent'")
.optional(),
due: tool.schema
.string()
.describe("Due date in YYYY-MM-DD format")
.optional(),
owner: tool.schema
.string()
.describe("Ticket owner name. If omitted, defaults to the git user.")
.optional(),
priority: tool.schema
.string()
.describe("Priority level, 1 (most urgent) to 5 (least). Left unset if omitted.")
.optional(),
},
async execute(args: Record<string, any>, context: ToolContext) {
if (!args.title) {
return "Error adding ticket: title is required";
}
const cmd: string[] = ["add", "--title", args.title];
if (args.description) {
descriptionLines(args.description).forEach((line: string) =>
cmd.push("--description", line),
);
}
// Blanks are dropped from these two on purpose: each value is one criterium or
// one step, so an empty line between them is spacing rather than content.
if (args.acceptance) {
itemLines(args.acceptance).forEach((line: string) =>
cmd.push("--acceptance", line),
);
}
if (args.plan) {
itemLines(args.plan).forEach((line: string) => cmd.push("--plan", line));
}
if (args.tags) cmd.push("--tags", args.tags);
if (args.due) cmd.push("--due", args.due);
if (args.owner) cmd.push("--owner", args.owner);
if (args.priority) cmd.push("--priority", String(args.priority));
try {
return await run(cmd, context);
} catch (err: any) {
return `Error adding ticket: ${err.message}`;
}
},
});
export const ls = tool({
description:
"List all worklane tickets, sorted by priority then title. Optionally filter by status and by priority.",
args: {
status: tool.schema
.enum(["backlog", "committed", "reviewing", "done"])
.describe("Filter by status")
.optional(),
priority: tool.schema
.string()
.describe("Only show tickets at this priority, 1 (most urgent) to 5 (least)")
.optional(),
},
async execute(args: Record<string, any>, context: ToolContext) {
const cmd = args.status ? ["ls", args.status] : ["ls"];
if (args.priority) cmd.push("--priority", String(args.priority));
try {
return await run(cmd, context);
} catch (err: any) {
// The directory is worth naming here: the usual failure is that there is
// no .worklane above it.
return `Error listing tickets: ${err.message}, Dir: ${resolveWorklaneDir(context)}`;
}
},
});
export const show = tool({
description:
"Show a complete ticket by UUID or title. Displays title, description, acceptance criteria, implementation plan, metadata, and comments.",
args: {
id: tool.schema
.string()
.describe("Ticket UUID prefix (4+ chars) or title substring (required)"),
},
async execute(args: Record<string, any>, context: ToolContext) {
if (!args.id) {
return "Error showing ticket: id is required";
}
try {
return await run(["show", args.id], context);
} catch (err: any) {
return `Error showing ticket: ${err.message}`;
}
},
});
export const mv = tool({
description:
"Move a ticket to a different status in the workflow (backlog -> committed -> reviewing -> done).",
args: {
ticket: tool.schema
.string()
.describe("Ticket UUID prefix (4+ chars) or title substring (required)"),
status: tool.schema
.enum(["backlog", "committed", "reviewing", "done"])
.describe("Target status (required)"),
force: tool.schema
.boolean()
.describe("Force move (allows non-sequential moves like done -> backlog). Use with caution.")
.optional(),
},
async execute(args: Record<string, any>, context: ToolContext) {
if (!args.ticket) {
return "Error moving ticket: ticket is required";
}
if (!args.status) {
return "Error moving ticket: status is required. Must be one of: backlog, committed, reviewing, done";
}
const cmd: string[] = ["mv", args.ticket, "--status", args.status];
if (args.force) cmd.push("--force");
try {
return await run(cmd, context);
} catch (err: any) {
return `Error moving ticket: ${err.message}`;
}
},
});
export const comment = tool({
description: "Add a comment to an existing ticket. The comment body supports markdown and newlines. Use it to add review feedback, notes, or status updates.",
args: {
ticket: tool.schema
.string()
.describe("Ticket UUID prefix (4+ chars) or title substring (required)"),
body: tool.schema
.string()
.describe("Comment body text (required). Supports markdown formatting and multiple paragraphs separated by newlines."),
owner: tool.schema
.string()
.describe("Comment author. If omitted, defaults to the git user.")
.optional(),
},
async execute(args: Record<string, any>, context: ToolContext) {
if (!args.ticket) {
return "Error adding comment: ticket is required";
}
if (!args.body) {
return "Error adding comment: body is required";
}
const cmd: string[] = ["comment", args.ticket, args.body];
if (args.owner) cmd.push("--owner", args.owner);
try {
return await run(cmd, context);
} catch (err: any) {
return `Error adding comment: ${err.message}`;
}
},
});
export const set = tool({
description:
"Replace fields on an existing worklane ticket. Only the fields given are touched; everything else in the file, including ticked criteria and handwritten sections, is left alone. Owner is not settable here, use chown. Status is not settable here, use mv. To add to a section instead of replacing it, use append.",
args: {
ticket: tool.schema
.string()
.describe("Ticket UUID prefix (4+ chars) or title substring (required)"),
title: tool.schema
.string()
.describe("New title. The file is renamed to match, keeping the UUID.")
.optional(),
description: tool.schema
.string()
.describe("New description, replacing the whole section. Use newlines between paragraphs.")
.optional(),
acceptance: tool.schema
.string()
.describe("New acceptance criteria, one per line, replacing the whole list.")
.optional(),
plan: tool.schema
.string()
.describe("New implementation plan steps, one per line, replacing the whole list.")
.optional(),
tags: tool.schema
.string()
.describe("New comma-separated tags, replacing the existing ones.")
.optional(),
due: tool.schema
.string()
.describe("New due date in YYYY-MM-DD format")
.optional(),
priority: tool.schema
.string()
.describe("New priority level, 1 (most urgent) to 5 (least)")
.optional(),
},
async execute(args: Record<string, any>, context: ToolContext) {
if (!args.ticket) {
return "Error setting fields: ticket is required";
}
const cmd: string[] = ["set", args.ticket];
if (args.title) cmd.push("--title", args.title);
if (args.description) {
descriptionLines(args.description).forEach((line: string) =>
cmd.push("--description", line),
);
}
if (args.acceptance) {
itemLines(args.acceptance).forEach((line: string) =>
cmd.push("--acceptance", line),
);
}
if (args.plan) {
itemLines(args.plan).forEach((line: string) => cmd.push("--plan", line));
}
if (args.tags) cmd.push("--tags", args.tags);
if (args.due) cmd.push("--due", args.due);
if (args.priority) cmd.push("--priority", String(args.priority));
if (cmd.length === 2) {
return "Error setting fields: give at least one field to change";
}
try {
return await run(cmd, context);
} catch (err: any) {
return `Error setting fields: ${err.message}`;
}
},
});
export const append = tool({
description:
"Add to a section of an existing worklane ticket without replacing what is there. The description gains a new paragraph; acceptance criteria and plan steps are added to the end of their lists, with steps numbered on from the last. Ticked criteria and handwritten content are left alone. To replace a section outright, use set.",
args: {
ticket: tool.schema
.string()
.describe("Ticket UUID prefix (4+ chars) or title substring (required)"),
description: tool.schema
.string()
.describe("Text to add as a new paragraph at the end of the description.")
.optional(),
acceptance: tool.schema
.string()
.describe("Acceptance criteria to add, one per line.")
.optional(),
plan: tool.schema
.string()
.describe("Implementation plan steps to add, one per line.")
.optional(),
},
async execute(args: Record<string, any>, context: ToolContext) {
if (!args.ticket) {
return "Error appending: ticket is required";
}
const cmd: string[] = ["append", args.ticket];
if (args.description) {
descriptionLines(args.description).forEach((line: string) =>
cmd.push("--description", line),
);
}
if (args.acceptance) {
itemLines(args.acceptance).forEach((line: string) =>
cmd.push("--acceptance", line),
);
}
if (args.plan) {
itemLines(args.plan).forEach((line: string) => cmd.push("--plan", line));
}
if (cmd.length === 2) {
return "Error appending: give at least one section to add to";
}
try {
return await run(cmd, context);
} catch (err: any) {
return `Error appending: ${err.message}`;
}
},
});
export const chown = tool({
description:
"Change the owner of a worklane ticket. This is the only writer of the Owner field: set does not offer it. The user is free text and is checked against nothing. Note the argument order, user first and ticket second, after Unix chown.",
args: {
user: tool.schema
.string()
.describe("New owner (required). Free text, checked against nothing."),
ticket: tool.schema
.string()
.describe("Ticket UUID prefix (4+ chars) or title substring (required)"),
},
async execute(args: Record<string, any>, context: ToolContext) {
if (!args.user) {
return "Error changing owner: user is required";
}
if (!args.ticket) {
return "Error changing owner: ticket is required";
}
try {
return await run(["chown", args.user, args.ticket], context);
} catch (err: any) {
return `Error changing owner: ${err.message}`;
}
},
});
/**
* check and uncheck differ only in the verb, so the tool is built once.
*
* The criteria are required here even though the CLI makes them optional: without any
* the CLI prompts, and there is no terminal on this side of the pipe. The CLI answers a
* missing criterion with the numbered list on stderr, which run() relays, so the model
* sees the options and can retry with a number.
*
* One value per line, like the other list arguments, and every value goes in one call.
* The CLI resolves them all before it writes, so a bad term costs nothing, and finishing
* a ticket is one tool call rather than one per box.
*/
function checkboxTool(verb: "check" | "uncheck") {
const past = verb === "check" ? "Checked" : "Unchecked";
return tool({
description:
`${past === "Checked" ? "Tick" : "Untick"} acceptance criteria on a worklane ticket. ` +
"Every criterion not named keeps its state. Identify each one by its number as shown by show, " +
"or by a substring of its text; a substring matching more than one is an error listing them. " +
`Pass several, one per line, to ${verb} them in a single call, or the single word all for every ` +
"criterion on the ticket. If any value cannot be resolved the whole call is refused and nothing " +
"is written, so a partly wrong list never leaves the ticket half-marked.",
args: {
ticket: tool.schema
.string()
.describe("Ticket UUID prefix (4+ chars) or title substring (required)"),
criteria: tool.schema
.string()
.describe(
"Criterion numbers as shown by show, or substrings of their text, one per line. " +
"Or the single word all for every criterion (required)",
),
},
async execute(args: Record<string, any>, context: ToolContext) {
if (!args.ticket) {
return `Error running ${verb}: ticket is required`;
}
const terms = args.criteria ? itemLines(String(args.criteria)) : [];
if (terms.length === 0) {
return `Error running ${verb}: criteria is required, give the numbers, substrings, or all`;
}
try {
return await run([verb, args.ticket, ...terms], context);
} catch (err: any) {
return `Error running ${verb}: ${err.message}`;
}
},
});
}
export const check = checkboxTool("check");
export const uncheck = checkboxTool("uncheck");