Menu
Skills in this repository
SkillsMP has collected 1,021 skills from protosphinx/sphinxstack. Open a skill to review its source and details.
protosphinx/sphinxstackShowing 40 of 1,021 collected skills.
Prevent cross-site request forgery with deliberate cookie behavior, request tokens, and origin verification. Use when browser credentials are attached automatically to state-changing requests.
Prevent reflected, stored, and DOM cross-site scripting through contextual output handling and safe APIs. Use when untrusted data can reach HTML, script, URL, style, or document execution contexts.
Work out an honest price and practice saying it. Cost floor, market check, the number said out loud without flinching. Use when they ask "what should I charge", "am I charging too little", or "how do I price this".
Process incoming webhooks with authentic verification, durable intake, deduplication, retries, and reconciliation. Use when an external provider sends at-least-once event notifications to an application.
Produce a livestream with defined program, signal chain, accessibility, moderation, redundancy, communications, and recording. Use when a real-time audience depends on coordinated live media delivery.
Coordinate and run an accessible live online session from rehearsal through follow-up. Use when speakers, media, captions, interaction, recording, moderation, consent, and technical fallback must work together in real time.
Produce an accessible video with integrated planning for captions, description, transcripts, readable visuals, understandable audio, and player behavior. Use when video must communicate to people with varied sensory, cognitive, motor, or language access needs.
Profile a financial dataset for structure, completeness, consistency, lineage, and reconciliation before analysis or modeling. Use when receiving ledger, transaction, billing, forecast, or reporting data.
Profile a Parquet file through metadata, schema, row groups, encodings, statistics, nulls, ranges, compression, and privacy-safe samples. Use when inspecting columnar data before analysis or ingestion.
Profile application performance with a representative workload, synchronized measurements, call-path evidence, and before-after verification. Use when latency, throughput, CPU, memory, I/O, or cost needs improvement.
Protect a public form with layered, accessible controls and measured false-positive handling. Use when spam or automation threatens submissions without justifying a hostile experience for everyone.
Prototype an interaction at the minimum fidelity needed to test behavior, state, timing, content, accessibility, and implementation risk. Use when a design question cannot be answered reliably by static screens.
Do a proper YouTube upload. Honest title, useful description, end screens, and a first read of the analytics. Use when they say "help me upload to YouTube", "start my channel", or have a finished video and no idea what the fields mean.
Evaluate a prospective customer against explicit problem, fit, authority, timing, and risk evidence, then choose a respectful next action. Use when deciding whether a lead should advance, nurture, refer, or close.
Interpret a balance sheet by verifying assets, liabilities, equity, liquidity signals, and changes since the prior reporting date. Use when reviewing financial position, working capital, debt, or management accounts.
Read a cash flow statement by tracing operating, investing, financing, noncash, and foreign-exchange movements back to balance-sheet change. Use when explaining why cash changed during a reporting period.
Read a package lockfile to identify resolved versions, dependency paths, integrity, sources, platforms, and unexpected changes. Use when reviewing installation, upgrade, vulnerability, or supply-chain evidence.
Read a stack trace by identifying the failure, call path, application frames, runtime context, and first evidence-backed investigation point. Use when an exception or crash includes nested call information.
Read an academic paper by tracing its question, claims, method, evidence, limitations, and relationship to other research. Use when studying, reviewing literature, or deciding whether a paper supports a claim.
Interpret an income statement by tracing revenue, costs, margins, and period changes back to their definitions and source evidence. Use when reviewing profit and loss results, budgets, forecasts, or management reports.
Read application logs as time-bounded evidence while preserving context, uncertainty, and sensitive-data boundaries. Use when a service symptom needs reconstruction from recorded events.
Rewrite the README so a visitor gets it in ten seconds. What it does, why, how to run it, what works today. Use when they say "fix my README", "make my repo look good", or a project they want to show has a bare README.
Reconcile bank activity to accounting records while identifying timing differences, missing entries, duplicates, fees, and unauthorized transactions. Use when closing a period or investigating a cash discrepancy.
Match a purchase order, receipt, invoice, and approval trail before payment or correction. Use when quantities, prices, taxes, delivery, or references do not agree.
Reconcile business expense records against accounts, receipts, categories, reimbursements, and the ledger, then produce a traceable exceptions report. Use when bookkeeping records do not match bank or card activity, receipts are missing, reimbursements are…
Reconcile intercompany balances and transactions bilaterally across entities, currencies, taxes, timing, and consolidation eliminations. Use when group accounts contain mismatched due-to, due-from, revenue, expense, loan, fee, or settlement positions.
Reconcile payment-processor transactions, fees, reserves, disputes, refunds, currencies, and settlement timing to payouts and bank deposits. Use when closing merchant cash or investigating settlement differences.
Reconcile source-system and warehouse records through stable keys, extraction boundaries, transformations, counts, values, aggregates, and exception ownership. Use when proving that analytical data represents the authoritative source population.
Preserve a work decision with context, evidence, authority, rationale, consequences, and revisit conditions. Use when future operators need to understand why a choice was made.
Get clean voice audio from the gear they already own. Room, mic distance, levels, and a cleanup pass in Audacity. Use when they say "my audio sounds bad", "how do I record my voice", or before any recording session.
Record usable room tone with matching microphones, gain, position, environment, and metadata for clean dialogue edits. Use when a production needs authentic ambient sound from the interview or scene location.
Recover a corrupted Git repository through preservation, object and reference diagnosis, alternate-copy inventory, minimal repair, and independent verification. Use when objects, refs, index, worktree, or repository metadata are damaged.
Diagnose and recover a stalled customer onboarding while preserving access, data, commitments, and trust. Use when a new customer cannot reach first value because setup, migration, integration, training, or ownership has failed.
Recover from a broken web deployment through user-impact containment, release and cache reconciliation, compatible rollback or forward repair, and independent journey verification. Use when mixed assets, service workers, APIs, schemas, flags, or edge state…
Recover from catastrophic source-control failure through evidence preservation, distributed object inventory, trust reconstruction, history repair, credential recovery, and independently verified restoration. Use when central repositories, mirrors, backups,…
Recover from a corrupted software release by containing untrusted artifacts, preserving evidence, rebuilding from verified sources, and restoring state-compatible service. Use when release integrity or provenance is uncertain.
Recover from cross-region data inconsistency through write fencing, immutable evidence, causal and business-state reconciliation, customer repair, safe authority restoration, and prevention. Use when partitions, replication, failover, clocks, or conflicting…
Recover from a failed data backfill through write containment, source-to-output lineage, affected-population reconciliation, versioned repair, downstream correction, and recurrence controls. Use when historical recomputation creates missing, duplicate,…
Recover from a global treasury control failure with independent containment, payment reconstruction, liquidity protection, obligation recovery, bank coordination, clean authority, and verified control restoration. Use when treasury instructions or balances…
Recover from a published media error by assessing harm, preserving versions, correcting the source and distribution graph, supporting affected people, and verifying propagation. Use when released audio, video, imagery, captions, graphics, or metadata contain…