Menu
SkillsMP has collected 34 skills from greenpau/agentx. Open a skill to review its source and details.
Showing 34 of 34 collected skills.
Audit or refresh AgentX implementation coverage, route reachability, contract traceability, diagram validity, and conformance evidence. Use after changing application source, AGENTS.md, an implementation skill, a Draw.io diagram, or a runtime boundary that…
Go coding standards and implementation directives for the terminal-first AgentX client. Use when creating, modifying, or reviewing Go code, runtime architecture, conformance status, package boundaries, configuration, constructors, errors, concurrency,…
Implement credential-source selection, OAuth and cloud-provider authentication, API client construction, proxy and TLS behavior, streaming and non-streaming requests, retries, fallback, and file transfer. Use when implementing the model/account network…
Implement conversation continuity across incremental persistence, resume, branching, rewind, compaction, context pressure, and crash recovery. Use when defining what survives a turn or process, rebuilding a session from durable events, or reducing context…
Implement noninteractive execution and the structured SDK stream, including provider-free native session management, CLI mode inference and validation, stdin acquisition, text/JSON/NDJSON output, the serialized headless turn runner, correlated control…
Implement the shared conversation turn engine, API-bound message projection, model streaming, retry and fallback, stop processing, derived progress/prompt assistance, speculative execution, and continuation decisions. Use when implementing or reviewing prompt…
Implement process bootstrap facts, observable application state, state-change reactions, working-directory identity, and model context assembly. Use when implementing session state, selectors, permission context storage, prompt precedence, cached and volatile…
Implement the append-only session transcript and recovery protocol, including native session inventory and deletion, message graph links, sidechains, metadata events, deduplication, write queues, branches, tombstones, compact and snip relinking, resume, fork,…
Develop, review, or extend AgentX across its complete application architecture. Use when tracing a requirement across runtime domains, deciding which implementation subskill owns a behavior, changing cross-domain boundaries, or auditing project coverage.
Implement user-invoked command discovery and execution together with prompt normalization, slash parsing, skill argument expansion, attachments, pasted content, hooks, and queued input priorities. Use when porting slash commands, input submission, command…
Implement context-pressure management and derived memory, including persistent file memory, relevance recall, shared team-memory synchronization, automatic dream consolidation, token estimation, automatic and manual compaction, session-memory compaction,…
Implement usage and cost accounting, privacy filtering, diagnostics, logs, traces, metrics, feature-evaluation evidence, health reporting, disk fallback, and update-check observation. Use when adding an event, sink, counter, profiler, diagnostic command, or…
Implement the observable product surfaces that adapt the shared session runtime to an interactive terminal, headless CLI, structured SDK stream, or feature-gated optional experience. Use when selecting a product entry surface, preserving cross-surface…
Implement portable filesystem, process, terminal, executable, lock, notification, sleep-prevention, updater-mechanism, signal, and shutdown behavior. Use when implementing an operating-system adapter, owning external resources, handling cancellation or exit,…
Implement or diagnose repo-local skill discovery from .codex/skills; skill invocation and isolation; prompt listing; and output-style discovery and prompt transformation. Use when skills are missing, unexpectedly unavailable, malformed, or rooted incorrectly,…
Create or revise repo-local skills connected by actionable routing statements from AGENTS.md through broad and increasingly specialized skills. Use when documenting AgentX engineering knowledge, adding or routing a skill, reorganizing the skill hierarchy, or…
Implement startup configuration, settings precedence, managed policy, remote policy refresh, settings synchronization, trust-sensitive derivation, and live reload. Use when implementing or auditing any behavior whose effective value depends on settings…
Implement permission modes, scoped rules, path and shell analysis, approval decisions, auto-mode safety, protected resources, and operating-system sandbox selection. Use when implementing or reviewing whether a capability may run, what must prompt, and which…
Create, review, or revise repository-compliant commit messages and commit-message files. Use when summarizing source changes for version control, choosing a change indicator, checking commit-message structure, or when the user asks to create a commit message…
Implement the capability boundary that turns model requests into validated, authorized, observable, and recoverable work. Use when tracing a tool call through registry resolution, permissions, hooks, sandboxing, execution, result normalization, or long-lived…
Implement execution that crosses process, worktree, machine, or service boundaries while preserving the shared session, capability, permission, task, and transcript contracts. Use for remote control, bridge sessions, reconnect and replay, subagents, teams,…
Implement the session-scoped extension plane that discovers, attributes, filters, names, merges, and invalidates commands, skills, output styles, plugins, hooks, MCP providers, and language servers. Use when deciding how extension contributions become…
Implement the interactive terminal session adapter: setup dialogs, REPL application state, guarded prompt dispatch, command queues, local and remote session adapters, cancellation, message projection, streaming display, fullscreen transcript, overlays, and…
Implement Model Context Protocol configuration, transport and connection lifecycle, OAuth, elicitation, channel permissions, result normalization, and plugin-provided language servers. Use when implementing external capability servers or editor-intelligence…
Implement delegated-agent and team execution while preserving agent-definition precedence, tool and permission filtering, explicit background-task durability boundaries, transcript ownership, resume behavior, worktree isolation, team identity, mailboxes,…
Implement operational services surrounding the semantic runtime, including credentials and providers, networking and proxies, portable filesystem/process/terminal ownership, cleanup, diagnostics, telemetry, usage, feature evaluation, and updates. Use when…
Implement optional or build-specific user experiences without weakening the core runtime, including persistent assistant/viewer mode, paged remote assistant history, voice push-to-talk and transcription, terminal companions, and supported external-build…
Implement plugin manifests, marketplaces, installation registries, dependency and policy checks, cache lifecycle, component loading, and the complete hook event protocol. Use when implementing extension packaging, plugin updates, or lifecycle interception…
Implement remote-control bridging and remote session placement while preserving session identity, worker leases, ordered replay, permission and control relays, reconnect behavior, and local safety authority. Use when implementing or reviewing…
Implement the shared session runtime from process startup through context composition and the recursive model loop. Use when defining runtime ownership, tracing a turn across bootstrap, state, and query boundaries, or deciding whether a core behavior belongs…
Implement asynchronous work and its distinct durability classes, including local shells, local and remote agents, teammates, workflows, monitors, main-session backgrounding, output storage, polling, notifications, cancellation, crash windows, recovery hints,…
Implement the retained-mode terminal renderer and interactive input stack, including layout, frame diffing, ANSI lifecycle, keyboard and mouse parsing, focus, selection, prompt editing, keybindings, Vim mode, paste handling, and durable prompt history. Use…
Implement the complete model-callable capability catalog, its deterministic registry assembly, per-tool semantic contracts, feature and profile gates, and output limits. Use when adding, porting, filtering, classifying, or testing a built-in, MCP, shell,…
Implement the common protocol for model-callable tools, including descriptors, aliases, schema and semantic validation, hook ordering, permission composition, scheduling, progress, cancellation, output persistence, and result pairing. Use when implementing or…