browser-forensics
Skill10 starsChanged 2 months agoApache-2.0View on GitHub
Inspect browser forensic recordings, control live browser DOM, execute agent workflows, pick elements with Ctrl+Shift+Click, capture screenshots, observe live mutations and interactions, perform DOM time-travel debugging, trace element lifecycles, and diagnose web apps via the browser-forensics MCP server (350 tools).
In this repository16 other files it gives its agents.
- browser-interactionSkill
- browser-state-recoverySkill
- dom-inspectionSkill
- dom-mutationSkill
- dom-targetingSkill
- extension-troubleshootingSkill
- human-like-interactionSkill
- mcp-troubleshootingSkill
- page-annotationSkill
- project-captureSkill
- project-reconstructionSkill
- regression-testingSkill
- selector-recoverySkill
- tab-managementSkill
- viewport-testingSkill
- visual-dom-comparisonSkill
The file
Files
SKILL.md
TeleDOM v4.1.0 — Temporal Browser Intelligence & Workflow Automation Skill (350 Tools)
MCP Tools Reference — TeleDOM Platform
350 certified MCP tools across 20 functional groups. Tool discovery tools: get_tool_groups, get_tool_catalog.
Tool Groups
session-forensics (21)
Historical forensic session management and analysis (recorded sessions, timelines, DOM states, diffs, lifecycle tracing).
list_sessions · get_session · export_session · import_session · delete_session · get_timeline · get_events · get_events_around · get_dom_state · get_dom_node · get_dom_subtree · diff_dom · trace_element · find_disappearing_elements · why_did_element_disappear · get_diagnostics · get_network_events · get_screenshots · annotate_session · get_annotations · get_recording_health
inspection (14)
Live page and element inspection: page metadata, element deep-info, visual state, DOM snapshots and analyzers.
inspect_live_page · inspect_live_element · get_element_visual_state · get_live_dom_snapshot · get_live_dom_subtree · get_tab_console_logs · get_tab_network_requests · get_element_ancestry · get_element_accessibility · get_computed_style · analyze_dom · search_dom · get_page_blueprint · get_element_fingerprint
targeting (3)
Element targeting resilience: TARGET generation, selector candidates with confidence, recovery, diagnostics.
generate_element_target · recover_selector · diagnose_selector_failure
interaction (16)
Page interaction: clicks, typing, hover, focus, keyboard, drag-and-drop, checkboxes, selects, waits.
interact_with_element · click_element · type_text · hover_element · focus_element · blur_element · press_keyboard_shortcut · scroll_to_element · scroll_page · drag_and_drop · set_input_checked · select_option · wait_for_condition · wait_for_dom_stable · set_interaction_profile · get_interaction_profile
tabs-browser (11)
Tab lifecycle and browser control: open/close/switch tabs, extension management, navigation, reload.
list_tabs · focus_tab · reload_tab · close_tab · open_tab · list_extensions · set_extension_enabled · toggle_extension · reload_extension · compare_extension_states · get_browser_session
selection-capture (5)
Interactive element selection (Ctrl+Shift+Click / picker) and element observation.
get_selected_element · start_element_picker · stop_element_picker · start_element_observation · stop_element_observation
viewport-responsive (5)
Viewport control and responsive testing: resize, presets, device emulation, multi-viewport workflows.
resize_viewport · reset_viewport · get_viewport_state · run_responsive_test · emulate_device
javascript (2)
Observable JavaScript execution with explicit outcome states and change capture.
execute_javascript · execute_js_and_capture_changes
dom-mutation (8)
First-class DOM mutation engine: operations with diff, transactions, undo/redo, history, preview.
mutate_dom · mutate_dom_transaction · undo_dom_mutation · redo_dom_mutation · get_mutation_history · preview_dom_mutation · preview_command · clone_dom_subtree
command-sequences (10)
Command sequences, recording, replay, import/export of deterministic command data.
execute_pipeline · execute_command_sequence · record_commands_start · record_commands_stop · list_command_recordings · get_command_recording · replay_command_recording · export_command_recording · import_command_recording · delete_command_recording
page-state (5)
Page state snapshots and time-travel comparison.
capture_page_state · compare_page_states · list_page_states · get_action_timeline · get_operation_trace
projects-knowledge (14)
Project folders, region capture/annotation, relationship graphs, reconstruction specs, agent packages.
create_page_project · list_projects · get_project · delete_project · capture_page_region · annotate_element · list_region_annotations · get_region_annotation · update_region_annotation · delete_region_annotation · get_region_relationship_graph · generate_reconstruction_spec · export_agent_package · import_project
screenshots (2)
Visual capture: page and element screenshots with geometry metadata.
capture_page_screenshot · capture_element_screenshot
security-privacy (2)
Capture redaction configuration and exclusion rules.
get_redaction_rules · set_redaction_rules
discovery (2)
Meta-tools: tool catalog and group discovery for agent self-orientation.
get_tool_catalog · get_tool_groups
Complete Tool List
| # | Tool | Group | Description |
| 1 | list_sessions | session-forensics | List all recorded browser forensic debugging sessions with metadata, timestamps, and stats. |
| 2 | get_session | session-forensics | Retrieve full metadata, capabilities, health status, and statistics for a specific debugging session. |
| 3 | export_session | session-forensics | Export a complete recording session as a portable, self-contained JSON bundle. |
| 4 | import_session | session-forensics | Import a recording session bundle from raw JSON string. |
| 5 | delete_session | session-forensics | Delete a recording session from storage. |
| 6 | get_timeline | session-forensics | Retrieve summary breakdown of events across the session timeline, including event categories and significant milestones. |
| 7 | get_events | session-forensics | Query recorded events with filtering by category (DOM, USER, ERROR, CONSOLE, NETWORK, etc.), type, timestamp range, target node, or search query. |
| 8 | get_events_around | session-forensics | Retrieve a focused contextual window of events occurring immediately before and after a specific timestamp or event ID. |
| 9 | get_dom_state | session-forensics | Reconstruct the complete DOM snapshot at an arbitrary timestamp or event ID using checkpoint delta replay. |
| 10 | get_dom_node | session-forensics | Inspect detailed properties of a specific DOM node at a given timestamp (tag, attributes, text, parent, children, visibility state). |
| 11 | get_dom_subtree | session-forensics | Reconstruct and extract the HTML of a specific subtree (e.g. #app or .gpt-panel) at a given timestamp. |
| 12 | diff_dom | session-forensics | Compare two DOM states between timestamp T1 and T2 (or event E1 and E2) and return structured additions, removals, moves, attribute, style, and text changes. |
| 13 | trace_element | session-forensics | Trace the entire chronological lifecycle of a DOM element from creation, mounting, mutations, style changes to unmounting/removal. |
| 14 | find_disappearing_elements | session-forensics | Automatically scan the session and identify all elements that existed temporarily and were subsequently removed or hidden. |
| 15 | why_did_element_disappear | session-forensics | Forensic root-cause diagnosis for why an injected or existing UI element disappeared. Pinpoints removal mechanism, ancestor container destruction, style changes, and correlated errors/network triggers. |
| 16 | get_diagnostics | session-forensics | Query recorded console messages, runtime errors, and unhandled promise rejections with stack traces. |
| 17 | get_network_events | session-forensics | Query recorded network requests and responses correlated with timing and duration. |
| 18 | get_screenshots | session-forensics | List visual checkpoints and screenshot checkpoints captured during the recording session. |
| 19 | annotate_session | session-forensics | Add an investigative annotation or hypothesis to the session timeline. |
| 20 | get_annotations | session-forensics | Retrieve all human and AI annotations created for a session. |
| 21 | get_recording_health | session-forensics | Run an automated integrity audit on a recording session to check sequence monotonicity, missing nodes, and capability health. |
| 22 | inspect_live_page | inspection | Inspect the current live browser page state, including URL, title, viewport dimensions, scroll positions, readyState, active and focused elements. |
| 23 | inspect_live_element | inspection | Deeply inspect a live DOM element on the active browser page by CSS selector, LogicalNodeId, or selectedElementRef, returning bounds, computed styles, visibility, attributes, state, role, aria, and parent context. |
| 24 | get_element_visual_state | inspection | Inspect detailed visual layout, occlusion, clipping, opacity, z-index, and viewport visibility for a live element. |
| 25 | get_live_dom_snapshot | inspection | Capture the current live virtual DOM state snapshot of the active or specified browser tab in HTML or structured JSON format. |
| 26 | get_live_dom_subtree | inspection | Reconstruct and extract the live HTML or node structure of a specific subtree on the active or specified browser tab. |
| 27 | get_tab_console_logs | inspection | Retrieve live intercepted console logs, uncaught JavaScript errors, and unhandled promise rejections for a specific tab or the active tab. |
| 28 | get_tab_network_requests | inspection | Retrieve live intercepted network requests and responses (Fetch, XHR) for a specific tab or the active tab, including method, URL, status, duration, and errors. |
| 29 | get_element_ancestry | inspection | Analyze element ancestry: ancestors chain (tag, selector, role, text, child index, sibling count, distance), nearby siblings (before/after with distance), and descendant summary (count, max depth, tags, interactive desce |
| 30 | get_element_accessibility | inspection | Extract accessibility metadata: explicit/implicit role, accessible name with sources, description, value, states, heading level, focusability, tabIndex, all aria-* attributes, and detected a11y issues. Input: target/sele |
| 31 | get_computed_style | inspection | Extract computed CSS style properties for an element. Input: target/selector + optional properties list (defaults to a layout-relevant set). Output: property→value map. Fails with STYLE_UNAVAILABLE when getComputedStyle |
| 32 | analyze_dom | inspection | Run a named DOM analyzer: analyze_forms, extract_links, analyze_media, get_css_variables, analyze_fonts, extract_color_palette, detect_zindex_conflicts, detect_layout_issues, census_interactive_elements, detect_semantic_ |
| 33 | search_dom | inspection | Search the DOM by text, tag and attribute patterns with scored results (selector, role, text, visibility, bounds). The fastest way to find elements without knowing selectors. Input: query (required), optional tag, attr, |
| 34 | get_page_blueprint | inspection | Generate a page blueprint: major sections with bounds and roles, hierarchy tree, key interactive elements, repeated components (patterns with occurrence counts), layout relationships and semantic regions. The architectur |
| 35 | get_element_fingerprint | inspection | Compute the structural DOM fingerprint of an element: stable hash, tag hierarchy, stable attributes, meaningful text, class list, role, dimensions, ancestor/descendant patterns, plus a volatility risk assessment with rea |
| 36 | generate_element_target | targeting | Build the canonical multi-strategy TARGET object for an element: ranked selector candidates with confidence, xpath, DOM path, text/attribute/structural fingerprints and bounds. Use before storing or acting on elements to |
| 37 | recover_selector | targeting | Recover a failed element selector via fingerprint matching: inspects the previous target snapshot, searches candidate matches, scores them, and attempts SAFE recovery (refuses below 0.62 confidence or ambiguous matches). |
| 38 | diagnose_selector_failure | targeting | Diagnose WHY a selector fails: syntax validity, match count, relaxation attempts that work, and human-readable diagnosis. Complements recover_selector (which attempts to fix). Input: selector. Output: validity, matches, |
| 39 | interact_with_element | interaction | Perform an interaction (click, double_click, right_click, hover, focus, blur, type, clear, press_key, select_option, scroll_into_view, scroll) on a live element and return before/after state and effect measurements. |
| 40 | click_element | interaction | Production-grade click with explicit mode: normal (synthetic pointer event sequence), double, right, or human-like (movement trajectory + click delay from the active interaction profile). Records which mode was actually |
| 41 | type_text | interaction | Robust typing into inputs, textareas and contenteditable with mode: append, replace (clear then type) or clear. Framework-sensitive: dispatches keydown/keypress/input/change per character. Input: target/selector, text, m |
| 42 | hover_element | interaction | Hover an element: pointerenter/mouseenter/mouseover/mousemove event sequence. Input: target/selector. Output: InteractionResult. |
| 43 | focus_element | interaction | Focus an element (native focus() + focus event). Input: target/selector. Output: InteractionResult. |
| 44 | blur_element | interaction | Blur (defocus) an element (native blur() + blur event). Input: target/selector. Output: InteractionResult. |
| 45 | press_keyboard_shortcut | interaction | Press a key or keyboard shortcut on an element (or the active element): keydown+keyup per key with modifier flags. Input: keys (e.g. ["Control","Shift","P"] or "Control+Shift+P"), optional target. Output: events fired an |
| 46 | scroll_to_element | interaction | Scroll an element into view (block: center). Input: target/selector. Output: before/after scroll positions and target selector. |
| 47 | scroll_page | interaction | Scroll the page by a distance: x/y pixel offsets. Input: x, y. Output: before/after scroll positions. |
| 48 | drag_and_drop | interaction | Drag and drop: HTML5 drag events (dragstart/dragenter/dragover/drop/dragend) plus pointer events between source and target (or by pixel offsets). Input: source (target specifier), optional target specifier or offsets. Ou |
| 49 | set_input_checked | interaction | Check/uncheck a checkbox or select a radio (peer radios with the same name are deselected). Dispatches input + change events. Input: target/selector, checked (default true). Output: checkedBefore/checkedAfter and events |
| 50 | select_option | interaction | Select an option in a dropdown: sets value and dispatches change. Input: target/selector, value. Output: InteractionResult. |
| 51 | wait_for_condition | interaction | Wait for a meaningful condition instead of arbitrary sleeps: dom_stable, selector_present, selector_visible, selector_absent, text_present, url_contains, element_count, readiness_state. Input: kind + condition params, ti |
| 52 | wait_for_dom_stable | interaction | Convenience wrapper for wait_for_condition {kind: dom_stable}: polls until two consecutive DOM length observations match. Input: timeoutMs. Output: satisfied + waitedMs. |
| 53 | set_interaction_profile | interaction | Set the human-interaction profile for subsequent interactions: DETERMINISTIC (zero delay), BALANCED (small natural delays), HUMAN_LIKE (realistic cadence, trajectories, hesitation) or CUSTOM (user timing parameters, seed |
| 54 | get_interaction_profile | interaction | Inspect the active interaction profile and the timing of the last action (requested mode, actual mode, per-phase timings, total duration). Output: InteractionProfileReport. |
| 55 | list_tabs | tabs-browser | List all open Chrome browser tabs across windows with tab ID, title, URL, active state, window ID, status, and recording status. |
| 56 | focus_tab | tabs-browser | Switch active focus to a specific browser tab by tabId and bring its window to the foreground. |
| 57 | reload_tab | tabs-browser | Reload a specific browser tab or the active tab with optional hard refresh (bypassCache). |
| 58 | close_tab | tabs-browser | Close a specific browser tab by tabId, URL substring, or pattern (e.g., "meet", "calendar"). |
| 59 | open_tab | tabs-browser | Open a new browser tab with the specified URL, meeting link, web page, or local file path in Chrome. |
| 60 | list_extensions | tabs-browser | List all installed Chrome extensions with ID, name, version, enabled status, installation type, and permissions. |
| 61 | set_extension_enabled | tabs-browser | Enable or disable a specific Chrome extension by ID (e.g. turn off extension to observe native clean UI, then turn back on). |
| 62 | toggle_extension | tabs-browser | Toggle the enabled status of a specific Chrome extension by ID. |
| 63 | reload_extension | tabs-browser | Reload an extension under development. If extensionId is provided, toggles and reloads that extension. If omitted, reloads the Forensic Recorder extension itself. |
| 64 | compare_extension_states | tabs-browser | Automatically perform a complete before/after comparative forensic audit: disables extension, reloads and captures clean native state/screenshot, enables extension, reloads and captures injected state/screenshot, and ret |
| 65 | get_browser_session | tabs-browser | Inspect the coherent browser session model: tabs with stable session identities, active tab, viewport state, extension state, snapshot count, command count, mutation history count, timeline event count, bound project id. |
| 66 | get_selected_element | selection-capture | Retrieve the DOM element visually selected by the user via Ctrl + Shift + Mouse Click in the live browser. |
| 67 | start_element_picker | selection-capture | Activate the interactive visual element picker mode in the live browser with hover highlighting and click selection. |
| 68 | stop_element_picker | selection-capture | Deactivate the visual element picker mode in the browser and restore normal cursor and interaction state. |
| 69 | start_element_observation | selection-capture | Start focused continuous recording and observation around a target element and its subtree/ancestors. |
| 70 | stop_element_observation | selection-capture | Stop focused element observation and assemble a complete correlation bundle with mutations, diagnostics, network activity, and root-cause analysis. |
| 71 | resize_viewport | viewport-responsive | Resize the viewport (width/height in pixels or a named preset). ALWAYS reversible: the original size is recorded on first use. Captures before/after page digests. Input: width+height, or preset (desktop-hd, tablet-ipad, |
| 72 | reset_viewport | viewport-responsive | Restore the original viewport size recorded before the first resize (guaranteed RESET_VIEWPORT semantics). Also clears the active preset/device. Input: none. Output: ViewportResizeResult with restored dimensions. |
| 73 | get_viewport_state | viewport-responsive | Inspect the viewport: current width/height/dpr/scroll plus the recorded original and whether it is currently modified. Input: none. |
| 74 | run_responsive_test | viewport-responsive | Run a multi-viewport responsive workflow: applies each size, records DOM length/interactive count/horizontal overflow per step, compares steps, then RESTORES the original viewport (unless restore:false). Input: sizes arr |
| 75 | emulate_device | viewport-responsive | Emulate a device profile: viewport + devicePixelRatio + touch metadata + reported UA (UA override applied only in a real browser session; honestly reported otherwise). Input: device (iphone-13, ipad-air, pixel-7, galaxy- |
| 76 | execute_javascript | javascript | Execute JavaScript with explicit outcome states: EXECUTED_SUCCESSFULLY, EXECUTED_WITH_ERROR, TIMED_OUT, SERIALIZATION_FAILED, BLOCKED_BY_CONTEXT, NOT_CONNECTED. Captures console output, duration, and DOM-change indicatio |
| 77 | execute_js_and_capture_changes | javascript | Composite: execute JavaScript AND capture the DOM state before/after with a structural diff summary + page-state snapshots. Same input as execute_javascript. Output: JSExecutionResult + before/after PageStateSnapshot + c |
| 78 | mutate_dom | dom-mutation | Apply a first-class DOM mutation with full observability (BEFORE → ACTION → AFTER → DIFF) and a guaranteed undo record. Operations: set_attribute, remove_attribute, set_text, replace_text, set_inner_html, set_outer_html, |
| 79 | mutate_dom_transaction | dom-mutation | Transactional DOM mutation: mode=begin opens a transaction, subsequent mutate_dom calls join it, mode=commit verifies and commits (all-or-nothing), mode=rollback reverts every step in reverse order. Input: mode, optional |
| 80 | undo_dom_mutation | dom-mutation | Undo the last DOM mutation (or the last mutation of the open transaction) using its immutable inverse record. Input: none. Output: success + undone mutation id. Idempotent-safe: reports "nothing to undo" when the stack i |
| 81 | redo_dom_mutation | dom-mutation | Redo the last undone DOM mutation (forward patch re-applied only when safe). Input: none. Output: success + redone mutation id. |
| 82 | get_mutation_history | dom-mutation | Inspect the DOM mutation history: entries (operation, target, summary, undo/redo flags), undo depth, redo depth, open transaction id. Input: limit (default 100). Side effect: none. |
| 83 | preview_dom_mutation | dom-mutation | Dry-run a mutation: validates the target, describes the expected change, counts affected nodes and warns about destructive operations — WITHOUT modifying anything. Input: identical to mutate_dom. Output: MutationPreview. |
| 84 | preview_command | dom-mutation | Dry-run preview of a DOM mutation: valid flag, expected change description, affected node count and warnings — WITHOUT applying anything. Input: operation + target + params (same as mutate_dom). Output: MutationPreview. |
| 85 | clone_dom_subtree | dom-mutation | Clone a DOM subtree (deep clone appended to a parent or the original parent; ids are never duplicated). Input: target, optional parent, copyAttributes. Output: DOMMutationResult (undoable). Side effect: adds cloned nodes |
| 86 | execute_pipeline | command-sequences | Execute a batch sequence of browser and extension actions in one call (e.g. reload extension, wait, reload tab, take screenshot directly to file, dump DOM to file) and aggregate all results. |
| 87 | execute_command_sequence | command-sequences | Execute a command sequence with per-command records (id, timestamp, target, args, status, result, duration, error): sequential execution, conditional continuation (condition.previousStepSucceeded), explicit stop-on-error |
| 88 | record_commands_start | command-sequences | Start recording subsequent tool invocations into a named command recording. Input: name, description, tags. Output: active recording metadata. Side effect: recording mode ON (adds latency-free capture to every tool call) |
| 89 | record_commands_stop | command-sequences | Stop the active command recording and persist it to storage (.mcpdom_recordings). Input: none (uses active recording). Output: the finished CommandRecording. |
| 90 | list_command_recordings | command-sequences | List saved command recordings with metadata (id, name, command count, timestamps, tags). Input: none. Side effect: none. |
| 91 | get_command_recording | command-sequences | Load one command recording in full (all commands with args and outcomes). Input: recordingId. Output: CommandRecording. Side effect: none. |
| 92 | replay_command_recording | command-sequences | Replay a saved command recording: executes each recorded tool call in order with deterministic arguments. Input: recordingId, stopOnError (default true). Output: CommandSequenceResult. Side effects: those of the recorded |
| 93 | export_command_recording | command-sequences | Export a command recording as portable JSON (with schema version) for another agent or session. Input: recordingId, outputPath (optional). Output: the exported JSON + save info. |
| 94 | import_command_recording | command-sequences | Import a command recording from JSON (previously exported). Input: recordingJson. Output: imported recording metadata. Side effect: writes to recordings storage. |
| 95 | delete_command_recording | command-sequences | Delete a saved command recording. Input: recordingId. Side effect: removes stored data (irreversible). |
| 96 | capture_page_state | page-state | Capture a page state snapshot (comparison anchor): url, title, viewport, dom length + hash, interactive count, extension state, pending mutations, annotation count. Input: none. Output: PageStateSnapshot (also recorded i |
| 97 | compare_page_states | page-state | Compare two page state snapshots ("what changed after this command?"): field-level diffs, DOM size delta, summary. Input: snapshotIdA + snapshotIdB (omit both to compare the two most recent snapshots). Side effect: none. |
| 98 | list_page_states | page-state | List captured page state snapshots (ids, timestamps, urls, dom sizes). Input: none. Side effect: none. |
| 99 | get_action_timeline | page-state | Query the chronological action timeline (TAB_OPENED, CLICKED, DOM_MUTATED, SNAPSHOT_CREATED, ERROR_OCCURRED, …) with optional filters (kind, sinceTimestamp, limit). The platform observability layer. Input: optional filte |
| 100 | get_operation_trace | page-state | Trace a single operation by its operationId: tool, start/end, duration, status, correlated timeline events, related error. Input: operationId (or latest). Side effect: none. |
| 101 | create_page_project | projects-knowledge | Create a portable page-analysis project folder (project.json, page.json, regions/, screenshots/, dom/, commands/, diffs/, metadata/, instructions/). The DOM snapshot is captured CLEANED: MCPDOM-injected artifacts exclude |
| 102 | list_projects | projects-knowledge | List page-analysis projects with manifests (names, page counts, region counts, timestamps). Input: none. Side effect: none. |
| 103 | get_project | projects-knowledge | Load a project in full: manifest, page manifest, and all region annotations. Input: projectName. Side effect: none. |
| 104 | delete_project | projects-knowledge | Delete a project folder permanently. Input: projectName. Side effect: irreversible file removal. |
| 105 | capture_page_region | projects-knowledge | Capture a page region into a project: resolves the element, captures LOCAL DOM + RELEVANT CONTEXT DOM (meaningful boundary), ranked selector candidates, fingerprint, styles, dimensions, auto-generated name and (optionall |
| 106 | annotate_element | projects-knowledge | Annotate a live element and capture it into a project (alias of capture_page_region with annotation semantics front and center): OBSERVED element data, USER comment/name/tags, INTENDED CHANGE and VERIFICATION CONDITIONS |
| 107 | list_region_annotations | projects-knowledge | List region annotations in a project (names, selectors, quality grades, intended changes). Input: projectName. Side effect: none. |
| 108 | get_region_annotation | projects-knowledge | Load one region annotation in full (observed facts, user fields, intended change, verification, quality score). Input: projectName, regionId. Side effect: none. |
| 109 | update_region_annotation | projects-knowledge | Update the USER fields of a region annotation (name, description, comment, tags, notes, intendedChange, verification). Observed facts are never editable. Quality is recomputed. Input: projectName, regionId, updates. Side |
| 110 | delete_region_annotation | projects-knowledge | Delete a region annotation from a project. Input: projectName, regionId. Side effect: removes the region file and updates manifests. |
| 111 | get_region_relationship_graph | projects-knowledge | Build the region relationship graph for a project: nodes (page + regions) and edges (contains, sibling-of, ancestor-of, overlaps) derived from live DOM containment. Input: projectName. Side effect: none. |
| 112 | generate_reconstruction_spec | projects-knowledge | Generate (and persist) the canonical page reconstruction specification for a project: metadata, viewport, structure, regions with selector candidates, hierarchy, semantic roles, visual constraints, interactions, selector |
| 113 | export_agent_package | projects-knowledge | Export a self-contained Agent handoff package for another AI agent: README.md, PROJECT.md, agent-instructions.md, project.json, pages/, regions/, snapshots/, diffs/, commands/, assets/, schemas/, verification/. The packa |
| 114 | import_project | projects-knowledge | Import a project previously exported via export_agent_package or a project folder copy: restores manifest, page, regions, diffs and command recordings into working storage. Input: projectDir. Side effect: copies files in |
| 115 | capture_page_screenshot | screenshots | Capture a screenshot of the visible browser page viewport with temporal, scroll, and viewport metadata. When outputPath is provided, saves decoded image directly to disk. |
| 116 | capture_element_screenshot | screenshots | Capture an element-specific screenshot bounded to the target element exact geometry and device pixel ratio. When outputPath is provided, saves decoded image directly to disk. |
| 117 | get_redaction_rules | security-privacy | Inspect the capture redaction configuration: all rules (key patterns, value patterns, attribute patterns) with enabled flags, and capture exclusions. Input: none. Side effect: none. |
| 118 | set_redaction_rules | security-privacy | Configure capture redaction: enable/disable existing rules, add custom key/value/attribute patterns, or add capture exclusions (selectors never captured). Built-in rules can be disabled but never removed. Input: enable/d |
| 119 | get_tool_catalog | discovery | Return the full MCP tool catalog with per-tool metadata: purpose, required context, accepted input, output, side effects, failure conditions, recovery strategy, and tool group. The meta-tool an agent calls FIRST to decid |
| 120 | get_tool_groups | discovery | List discoverable tool groups (inspection, targeting, interaction, viewport, javascript, mutation, sequences, session, projects, security, discovery) with descriptions and member tool names. Input: none. Side effect: non |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.