@staff0rd/assist 0.654.1 → 0.656.0

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@staff0rd/assist",
-	"version": "0.654.1",
+	"version": "0.656.0",
 	"type": "module",
 	"main": "dist/index.js",
 	"bin": {

package/pi/advice.ts

@@ -0,0 +1,24 @@
+import { spawnSync } from "node:child_process";
+import type { ExtensionAPI } from "@earendil-works/pi-coding-agent";
+
+function compose(cwd: string): string {
+	try {
+		const result = spawnSync("assist", ["advise"], { cwd, encoding: "utf8" });
+		return result.stdout?.trim() ?? "";
+	} catch {
+		return "";
+	}
+}
+
+export default function (pi: ExtensionAPI) {
+	let advice = "";
+
+	pi.on("session_start", (_event, ctx) => {
+		advice = compose(ctx.cwd);
+	});
+
+	pi.on("before_agent_start", (event) => {
+		if (!advice || event.systemPrompt.includes(advice)) return undefined;
+		return { systemPrompt: `${event.systemPrompt}\n\n${advice}` };
+	});
+}

package/claude/CLAUDE.md

@@ -1,45 +0,0 @@
-After any code change, run `/verify` to ensure all checks pass. Run it bare — never pipe it through `head`, `tail`, `grep`, `rg` or `wc`, as it already prints only what failed.
-
-`assist` is installed globally. Use it directly (e.g., `assist commit "message"`).
-
-## Editing files
-
-Always edit files with the Write and Edit tools. Never edit a file by shelling out — no `sed -i`, no heredoc rewrites, no Python or Node scripts that write the file. This overrides any session instruction that says to prefer Bash for file changes.
-
-## Writing markdown
-
-Do not hard-wrap markdown. One line per paragraph, per bullet, per heading — let it soft-wrap. Never break a line at some column width.
-
-When renaming TypeScript files or symbols, use the refactor commands instead of doing it manually:
-
-- `assist refactor rename file <source> <destination>` — rename/move a file and update all imports
-- `assist refactor rename symbol <file> <oldName> <newName>` — rename a variable, function, class, or type across the project
-- `assist refactor extract <file> <functionName> <destination>` — extract a function and its private dependencies to a new file
-  All default to dry-run; add `--apply` to execute.
-  When using extract, name the destination file after the exported function (e.g. `updateWorkerCapacity.ts` for `updateWorkerCapacity`) to satisfy the `oxlint-rules/filenameConvention.ts` lint rule.
-
-Do not modify `claude/settings.json` without asking the user first. Only read-only commands should be added to the allow list — write operations (add, remove, set, delete) must require confirmation.
-
-## Fetching Jira context
-
-When the user mentions a Jira issue key (e.g. `BAD-671`, `PROJ-123`), use the Atlassian MCP to fetch context.
-
-## Editing Jira issues with Smart Links
-
-When editing a Jira issue whose description (or comment) contains Smart Links — rendered issue-key/URL chips, stored as ADF `inlineCard` nodes — edit via `editJiraIssue` with `contentFormat: "adf"` and emit real `inlineCard` nodes. Never round-trip such content through `contentFormat: "markdown"`: markdown has no representation for `inlineCard`, so every existing Smart Link is flattened to a plain browse URL and lost. Jira does not auto-promote plain browse URLs back into Smart Links, so the damage is not self-healing. Inspect the current stored ADF first and preserve existing `inlineCard` nodes when making a small edit. For anything beyond a trivial edit, build the ADF programmatically (a small script) rather than hand-authoring it inline — full-description ADF is verbose and easy to get wrong.
-
-## Drafting messages to send on the user's behalf
-
-When drafting a message to send on the user's behalf (Slack, PR, email, comment): state only the confirmed conclusion and the facts behind it. No hedging, no speculative next-steps, no fixes not agreed to, no pleasantries. Match the scope of what was asked — nothing added.
-
-## Commenting code
-
-Do not include comments for standard logic or syntax. Only comment if the specific line involves unintuitive complexity or a hack.
-
-## Backlog item IDs
-
-A bare `a` followed by hex (e.g. `a791`, `a3f0`) is an assist backlog item ID, not a git SHA. Read it with `assist backlog view <id>`.
-
-## Backlog items, PRs, and commits
-
-Do not mention assist backlog items in PR descriptions or PR comments. The reference goes the other way: mention the relevant commits or PRs in the assist backlog item.