@gajae-code/coding-agent 0.8.2 → 0.9.0

package/src/prompts/tools/rewind.md

@@ -6,7 +6,7 @@ Requirements:
 - `report` is REQUIRED and must be concise, factual, and actionable.
 - Include key findings, decisions, and any unresolved risks.
 - Do not include raw scratch logs unless essential.
-- You MUST call this before yielding if a checkpoint is active.
+- `checkpoint`'s must-rewind-before-yield rule applies: never yield with a checkpoint still active.
 
 Behavior:
 - If no checkpoint is active, this tool errors.

package/src/prompts/tools/search-tool-bm25.md

@@ -5,7 +5,7 @@ Activate hidden tools (MCP and built-in) when you need a capability not in your
 {{#if discoverableMCPToolCount}}Total discoverable tools available: {{discoverableMCPToolCount}}.{{/if}}
 Input:
 - `query` — required natural-language or keyword query
-- `limit` — optional maximum number of tools to return and activate (default `8`)
+- `limit` — optional maximum number of tools to return and activate (default `8`; start with 5–10 if unsure)
 
 Behavior:
 - Searches hidden tool metadata using BM25-style relevance ranking
@@ -14,16 +14,6 @@ Behavior:
 - Repeated searches add to the active tool set; they do not remove earlier selections
 - Newly activated tools become available before the next model call in the same overall turn
 
-Notes:
-Start with `limit` 5–10 if unsure.
-- `query` is matched against tool metadata fields:
-  - `name`
-  - `label`
-  - `server_name` (MCP tools)
-  - `mcp_tool_name` (MCP tools)
-  - `description` / `summary`
-  - input schema property keys (`schema_keys`)
-
 Not for repository/file/code search. Tool discovery only.
 
 Returns JSON with:

package/src/prompts/tools/search.md

@@ -2,7 +2,7 @@ Searches files using powerful regex matching.
 
 <instruction>
 - Supports Rust regex syntax (RE2-style — no lookaround or backreferences). Use line anchors or post-filters instead of (?!…)/(?<!…)
-- `paths` is required and accepts an array of files, directories, globs, or internal URLs
+- `paths` accepts an array of files, directories, globs, or internal URLs; when omitted, the whole working directory is searched
 - `paths` is an array; do not embed commas or spaces inside a single entry. Pass `["src", "tests"]` not `["src,tests"]`.
 - Cross-line patterns are detected from literal `\n` or escaped `\\n` in `pattern`
 </instruction>

package/src/prompts/tools/skill.md

@@ -8,14 +8,15 @@ Invoke another available skill in the current turn.
 <instruction>
 - `name` is the skill name as it appears in `/skill:<name>` (e.g. `ralplan`, `ultragoal`, `team`, `deep-interview`)
 - `args` is the free-form argument string the skill would receive after `/skill:<name>` on the command line
-- The skill tool dispatches the callee's SKILL.md as a user-attribution custom message in the current turn (steering the stream when active, appending otherwise). Before dispatch, the tool atomically demotes the caller and promotes the callee in `.gjc/state/` by calling `gjc state <caller> handoff --to <callee>` in-process.
-- The chain is refused unless the caller's `current_phase` is in `{complete, completed, handoff, failed, cancelled, canceled, inactive}`. To prepare the active skill for chaining, write `current_phase: "handoff"` to its mode-state via `gjc state <skill> write --input '{"current_phase":"handoff"}' --json`. The skill tool itself then runs `gjc state <skill> handoff --to <callee>` in-process to atomically demote the caller and promote the callee — you do not need to run the handoff verb separately.
+- The tool loads the callee's SKILL.md into the current turn and handles the caller→callee state handoff itself.
+- The chain is refused while the caller skill is still active. If your current skill has not yet reached a terminal phase, prepare it first with `gjc state <skill> write --input '{"current_phase":"handoff"}' --json`; no other handoff command is needed.
 - Call once per chain step. To chain `A → B → C`, A calls `skill(B)`; B's next agent turn calls `skill(C)`.
 </instruction>
 
 <critical>
 - Do NOT use this tool to "remind yourself" of a skill you're already running. The current SKILL.md is already in your context.
 - Do NOT chain into the same skill recursively. If a skill's flow needs another iteration, follow its in-document instructions.
+- `name` MUST be one concrete skill name, NOT a glob or wildcard. Passing `*`, `?`, or a pattern like `git-*` is rejected immediately — the `--skills '*'` launch filter is unrelated to this tool's `name`.
 - The chained skill's planning/execution-boundary rules still apply. Chaining does not grant execution approval.
 </critical>
 

package/src/prompts/tools/ssh.md

@@ -1,7 +1,8 @@
 Runs commands on remote hosts.
 
 <instruction>
-You MUST build commands from the reference below
+You MUST build commands from the reference below.
+The local coreutils restrictions (`cat`/`grep`/`find`/`head`/`tail` bans) do NOT apply on remote hosts — `read`/`search`/`find` cannot reach them, so these shell commands are the only tools available there.
 </instruction>
 
 <commands>

package/src/prompts/tools/subagent.md

@@ -2,12 +2,12 @@ Lists, inspects, awaits, pauses, resumes, steers, or cancels detached task subag
 
 Task launches return immediately. Use this tool when you need direct control over those running subagents. Prefer `subagent` for task subagents; generic `job` remains available for non-subagent jobs and compatibility fallback access.
 
-`verbosity` controls output size: `receipt` (default) returns status metadata plus a single <=280-character result/error preview and an `agent://<id>` output ref when available; `preview` returns <=2000 characters; `full` returns <=12000 characters and requires explicit `ids`.
+`verbosity` controls output size: `receipt` (default) returns status metadata plus a single ≤280-character result/error preview and an `agent://<id>` output ref when available; `preview` returns ≤2000 characters; `full` returns ≤12000 characters and requires explicit `ids`.
 
 # Operations
 
 ## `action: "list"`
-Snapshot your visible detached subagents, including `running`, `paused`, `queued`, and terminal subagents when retained. Output is receipt-only by default; use `verbosity: "preview"` for a bounded preview or inspect explicit `ids` with `verbosity: "full"` when fuller retained text is necessary.
+Snapshot your visible detached subagents, including `running`, `paused`, `queued`, and terminal subagents when retained. Optional `limit` (1–50) caps how many subagents are returned. Output is receipt-only by default; use `verbosity: "preview"` for a bounded preview or inspect explicit `ids` with `verbosity: "full"` when fuller retained text is necessary.
 
 ## `action: "inspect"`
 Inspect selected subagents by `ids`; omit `ids` to inspect current running subagents. Terminal subagents return receipt-only output by default, with an `agent://<id>` ref when a verified output artifact is available. `verbosity: "full"` requires explicit `ids`.

package/src/prompts/tools/task.md

@@ -3,12 +3,11 @@ Launches subagents to parallelize workflows.
 - Results are delivered automatically when complete.
 - The tool result lists the assigned task ids (e.g. `0-AuthLoader`) — those are the live agent ids.
 {{#if ircEnabled}}
-- Coordinate with running tasks via `irc` using those ids. `subagent` cancel terminates a task and **cannot carry a message** — never cancel because an await timed out; cancel only when the task has actually failed, gone off-track, or become unrecoverably wrong.
-- If genuinely blocked on completion, wait with `subagent` action `await` and a timeout; timeout only bounds your wait and does not stop or fail the subagent.
+- Coordinate with running tasks via `irc` using those ids. `subagent` cancel terminates a task and **cannot carry a message**.
 {{else}}
-- If genuinely blocked on completion, wait with `subagent` action `await` and a timeout; timeout only bounds your wait and does not stop or fail the subagent.
-- Use `subagent` action `inspect` or `list` to snapshot manager state; `cancel` only when a task has actually failed, gone off-track, or become unrecoverably wrong.
+- Use `subagent` action `inspect` or `list` to snapshot manager state.
 {{/if}}
+- To wait or cancel, use the `subagent` tool; its await/cancel doctrine applies (never cancel just because an await timed out).
 
 {{#if ircEnabled}}
 Subagents have no conversation history, but they can reach you and their siblings live via the `irc` tool. Front-load every fact, file path, and direction they need in {{#if contextEnabled}}`context` or `assignment`{{else}}each `assignment`{{/if}}.
@@ -19,7 +18,7 @@ Subagents have no conversation history. Every fact, file path, and direction the
 <parameters>
 - `agent`: agent type for all tasks
 - `tasks`: tasks to execute in parallel
- - `.id`: CamelCase, ≤32 chars
+ - `.id`: filesystem-safe, ≤48 chars, matching `[A-Za-z0-9][A-Za-z0-9_-]*`; prefer CamelCase
  - `.description`: UI label only — subagent never sees it
  - `.assignment`: complete self-contained instructions; one-liners and missing acceptance criteria are PROHIBITED
 {{#if contextEnabled}}- `context`: shared background prepended to every assignment; session-specific only{{/if}}

package/src/prompts/tools/todo-write.md

@@ -12,7 +12,7 @@ Allowed `op` values are only `init`, `start`, `done`, `drop`, `rm`, `append`, an
 |`start`|`task`|Mark in progress|
 |`done`|`task` or `phase`|Mark completed|
 |`drop`|`task` or `phase`|Mark abandoned|
-|`rm`|`task` or `phase`|Remove|
+|`rm`|`task` or `phase` or *(none = clear all tasks)*|Remove a task; with `phase`, empties the phase (the phase entry remains); bare `rm` clears every task in every phase|
 |`append`|`phase`, `items: string[]`|Append tasks to `phase`; lazily creates phase|
 |`note`|`task`, `text`|Append a note to a task. Reminders for future-you only.|
 

package/src/prompts/tools/vim.md

@@ -1,6 +1,7 @@
 Vim-style `edit` mode. The tool name stays `edit`; every call requires `file`, and the buffer loads automatically on first use.
 - `{"file": "path"}` - view file
 - `{"file": "path", "steps": [{"kbd": ["…"], "insert": "…"}]}` - edit file
+- `pause` (optional boolean) - a "paused" call skips the auto-save after its steps and may leave the last step in INSERT mode; the buffer stays open for the next call. Non-paused calls auto-save once after all steps complete.
 
 **Multi-location edits: always edit highest line number first (bottom-up).** Each insert shifts lines below it.
 

package/src/prompts/tools/web-search.md

@@ -3,9 +3,11 @@ Searches the web for up-to-date information beyond knowledge cutoff.
 <instruction>
 - You SHOULD prefer primary sources (papers, official docs) and corroborate key claims with multiple sources
 - You MUST include links for cited sources in the final response
+- Provider-neutral params: `recency` (freshness window), `limit`/`num_search_results` (result counts), `max_tokens`, `temperature`
 </instruction>
 
 <xai>
+The parameters below apply ONLY when the active search provider is `xai`; ignore them (and never pass them) on any other provider.
 - With provider `xai`, use `xai_search_mode: "web"` for normal web search, `"x"` for X/Twitter search, or `"web_and_x"` when both surfaces are relevant.
 - xAI web filters: `allowed_domains` or `excluded_domains` (max 5, mutually exclusive), plus `enable_image_understanding` and `enable_image_search`.
 - xAI X filters: `allowed_x_handles` or `excluded_x_handles` (max 20, mutually exclusive), `from_date`, `to_date`, `enable_image_understanding`, and `enable_video_understanding`.

package/src/prompts/tools/write.md

@@ -3,10 +3,16 @@ Creates or overwrites file at specified path.
 <conditions>
 - Creating new files explicitly required by task
 - Replacing entire file contents when editing would be more complex
-- Supports `.tar`, `.tar.gz`, `.tgz`, and `.zip` archive entries via `archive.ext:path/inside/archive`
-- Supports SQLite row operations via `db.sqlite:table` (insert), `db.sqlite:table:key` (update with JSON content, delete with empty content)
 </conditions>
 
+<instruction>
+- Archives: write entries inside `.tar`, `.tar.gz`, `.tgz`, and `.zip` via `archive.ext:path/inside/archive`.
+- SQLite rows:
+  - `db.sqlite:table` with JSON content — insert a row
+  - `db.sqlite:table:key` with JSON content — update the row with that primary key
+  - `db.sqlite:table:key` with empty content — DELETE that row (destructive; double-check the key)
+</instruction>
+
 <critical>
 - You SHOULD use Edit tool for modifying existing files (more precise, preserves formatting)
 - You NEVER create documentation files (*.md, README) unless explicitly requested

package/src/sdk.ts

@@ -4,7 +4,7 @@ import {
 	type AgentMessage,
 	type AgentTelemetryConfig,
 	type AgentTool,
-	AppendOnlyContextManager,
+	type AppendOnlyContextManager,
 	INTENT_FIELD,
 	type ThinkingLevel,
 } from "@gajae-code/agent-core";
@@ -31,7 +31,11 @@ import {
 	prompt,
 	Snowflake,
 } from "@gajae-code/utils";
-
+import {
+	createAppendOnlyContextManager,
+	providerSupportsAppendOnlyAuto,
+	resolveAppendOnlyMode,
+} from "./append-only-mode";
 import { type AsyncJob, AsyncJobManager, isBackgroundJobSupportEnabled, jobElapsedMs } from "./async";
 import { loadCapability } from "./capability";
 import { type Rule, ruleCapability, setActiveRules } from "./capability/rule";
@@ -615,23 +619,12 @@ function registerJsVmCleanup(): void {
 	postmortem.register("js-vm-cleanup", disposeAllVmContexts);
 }
 
-/**
- * Resolve whether to enable append-only context mode based on the setting and provider.
- *
- * - `"on"` → always enable
- * - `"off"` → never enable
- * - `"auto"` → enable for DeepSeek (prefix-caching provider)
+/*
+ * Append-only context-mode resolution + manager construction live in
+ * ./append-only-mode so the initial build, the runtime model/setting-change
+ * path, and the status UI share one implementation. Re-exported for importers/tests.
  */
-function resolveAppendOnlyMode(setting: "auto" | "on" | "off" | undefined, provider: string): boolean {
-	switch (setting ?? "auto") {
-		case "on":
-			return true;
-		case "off":
-			return false;
-		default:
-			return provider === "deepseek";
-	}
-}
+export { createAppendOnlyContextManager, providerSupportsAppendOnlyAuto, resolveAppendOnlyMode };
 
 function customToolToDefinition(tool: CustomTool): ToolDefinition {
 	const definition: ToolDefinition & { [TOOL_DEFINITION_MARKER]: true } = {
@@ -866,6 +859,10 @@ function withEmbeddedDefaultGjcSkills(skills: Skill[]): Skill[] {
 	return [...byName.values()];
 }
 
+export function resolveIntentTracingEnabled(intentTracingSetting: boolean | undefined, hasUI: boolean): boolean {
+	return (!!intentTracingSetting || $flag("PI_INTENT_TRACING")) && hasUI;
+}
+
 export async function createAgentSession(options: CreateAgentSessionOptions = {}): Promise<CreateAgentSessionResult> {
 	const cwd = options.cwd ?? getProjectDir();
 	const agentDir = options.agentDir ?? getDefaultAgentDir();
@@ -1125,8 +1122,15 @@ export async function createAgentSession(options: CreateAgentSessionOptions = {}
 				rulebookRules.push(rule);
 			}
 		}
-		if (existingSession.injectedTtsrRules.length > 0) {
-			ttsrManager.restoreInjected(existingSession.injectedTtsrRules);
+		if (ttsrManager.getSettings().enabled !== false) {
+			if ((existingSession.ttsrMessageCount ?? 0) > 0) {
+				ttsrManager.restoreMessageCount(existingSession.ttsrMessageCount ?? 0);
+			}
+			if (existingSession.injectedTtsrRuleRecords && existingSession.injectedTtsrRuleRecords.length > 0) {
+				ttsrManager.restoreInjected(existingSession.injectedTtsrRuleRecords);
+			} else if (existingSession.injectedTtsrRules.length > 0) {
+				ttsrManager.restoreInjected(existingSession.injectedTtsrRules);
+			}
 		}
 		return { ttsrManager, rulebookRules, alwaysApplyRules };
 	});
@@ -1765,7 +1769,11 @@ export async function createAgentSession(options: CreateAgentSessionOptions = {}
 
 		const repeatToolDescriptions = settings.get("repeatToolDescriptions");
 		const eagerTasks = settings.get("task.eager");
-		const intentField = settings.get("tools.intentTracing") || $flag("PI_INTENT_TRACING") ? INTENT_FIELD : undefined;
+		const intentTracingEnabled = resolveIntentTracingEnabled(
+			settings.get("tools.intentTracing"),
+			options.hasUI ?? false,
+		);
+		const intentField = intentTracingEnabled ? INTENT_FIELD : undefined;
 		const rebuildSystemPrompt = async (
 			toolNames: string[],
 			tools: Map<string, AgentTool>,
@@ -1837,6 +1845,7 @@ export async function createAgentSession(options: CreateAgentSessionOptions = {}
 				eagerTasks,
 				secretsEnabled,
 				workspaceTree: workspaceTreePromise,
+				subagent: options.parentTaskPrefix !== undefined,
 			});
 
 			if (options.systemPrompt === undefined) {
@@ -2045,7 +2054,7 @@ export async function createAgentSession(options: CreateAgentSessionOptions = {}
 
 		const appendOnlyContext =
 			model && resolveAppendOnlyMode(settings.get("provider.appendOnlyContext"), model.provider)
-				? new AppendOnlyContextManager()
+				? createAppendOnlyContextManager(model.provider)
 				: undefined;
 		if (appendOnlyContext && options.forkContextSeed && !hasExistingSession) {
 			if (options.forkContextSeed.appendOnlyPrefixSnapshot) {
@@ -2216,6 +2225,7 @@ export async function createAgentSession(options: CreateAgentSessionOptions = {}
 			onResponse,
 			convertToLlm: convertToLlmFinal,
 			rebuildSystemPrompt,
+			workspaceTree: resolvedWorkspaceTree,
 			reloadSshTool,
 			requestedToolNames: requestedToolNameSet,
 			discoverableToolAllowedNames: options.discoverableToolAllowedNames,